Local bilingual text generation, research, fine-tuning, offline applications, and resource-conscious deployment
Context
4K
Reasoning
5/10
Speed
7/10
Status
Available open-weight model; older first-generation Yi base model
Long-document completion, local deployment, English-Chinese text generation, research, and downstream fine-tuning
Context
200K
Reasoning
4/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Available as downloadable open weights; legacy-generation model
Input
No official hosted API price verified; downloadable weights
Output
No official hosted API price verified; downloadable weights
Local bilingual English-Chinese chat, personal projects, academic experimentation, and fine-tuning on modest open-weight infrastructure
Context
4K
Reasoning
5/10
Speed
7/10
Status
Open-weight and downloadable; legacy historical model
Input
No official hosted API price verified; self-hosted weights have no per-token provider charge
Output
No official hosted API price verified; self-hosted weights have no per-token provider charge
Local text generation, code completion, mathematics, bilingual English-Chinese applications, research, and downstream fine-tuning
Context
4K
Reasoning
6/10
Speed
7/10
Status
Available as an open-weight downloadable model; older Yi-generation model with no verified first-party hosted API offering for this exact model
Long-context document processing, code generation, mathematics, bilingual English-Chinese text generation, local inference, and domain fine-tuning
Context
200K
Reasoning
5/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight base model; legacy generation but still publicly downloadable
Input
No official hosted API price; downloadable weights
Output
No official hosted API price; downloadable weights
Self-hosted bilingual text generation, research, custom fine-tuning, coding experiments, and English-Chinese applications
Context
4K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model; older-generation base checkpoint
Input
No official hosted API price verified; self-hosted weights
Output
No official hosted API price verified; self-hosted weights
Long-document analysis, English-Chinese generation, retrieval experiments, research, and self-hosted fine-tuning
Context
200K
Reasoning
7/10
Speed
4/10
Status
Available as downloadable open-weight model; no official hosted API availability or retirement date verified
Self-hosted bilingual assistants, English-Chinese dialogue, open-weight LLM research, private inference, quantization, and fine-tuning
Context
4K
Reasoning
6/10
Speed
3/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight; downloadable; legacy-generation model with no verified provider-managed hosted API availability
Long-context chat, complex text analysis, multilingual generation, prediction, and general-purpose enterprise language applications.
Context
32K
Reasoning
7/10
Speed
6/10
Status
Available through 01.AI API documentation; current lifecycle details are not explicitly stated by the provider.
Input
$3 per 1 million input tokens
Output
$3 per 1 million output tokens
Cost-sensitive hosted chat, Chinese-English generation, coding, mathematics, reasoning, summarization and high-volume API workloads
Context
16K
Reasoning
8/10
Speed
9/10
Status
Proprietary hosted API model; current public availability and lifecycle status are not clearly documented
Input
$0.14 per 1 million input tokens historically reported; verify current pricing with 01.AI
Output
$0.14 per 1 million output tokens historically reported; verify current pricing with 01.AI
Local text generation, self-hosted applications, experimentation, domain adaptation, and fine-tuning
Context
4K
Reasoning
5/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight downloadable model for self-hosted and local inference; no hosted inference provider is currently listed on its Hugging Face model page.
Local conversational applications, instruction following, lightweight coding, bilingual English-Chinese text generation, experimentation, and self-hosted inference
Context
4K
Reasoning
4/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight; accessible for local deployment; no explicit retirement date found
Local text generation, research, fine-tuning, Chinese-English applications, and cost-sensitive self-hosted deployments
Context
4K
Reasoning
6/10
Speed
7/10
Status
Open-weight and downloadable; no official deprecation or shutdown date found
Input
No official first-party hosted price; downloadable weights under Apache 2.0
Output
No official first-party hosted price; downloadable weights under Apache 2.0
Self-hosted conversational assistants, local text generation, lightweight coding help, multilingual experimentation, and fine-tuning research.
Context
4K
Reasoning
6/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model; official weights remain accessible, with no exact provider-published deprecation or shutdown date verified.
Input
No official hosted API price verified; self-hosted weights are available under Apache 2.0.
Output
No official hosted API price verified; self-hosted weights are available under Apache 2.0.
Local bilingual text generation, research, domain adaptation, and fine-tuning
Context
4K
Reasoning
7/10
Speed
4/10
Status
Open-weight and currently downloadable; standard 4K-context base checkpoint
Self-hosted bilingual assistants, English-Chinese text generation, coding support, mathematics, reasoning, fine-tuning, and privacy-sensitive deployments
Context
4K
Reasoning
7/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight model remains available for self-hosted deployment; 01.AI hosted model-platform API service ended on 2026-09-03
Local code completion, code generation, multilingual programming tasks, long-context source-code analysis, and downstream fine-tuning
Context
131K
Reasoning
3/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available as downloadable open-weight model; no exact first-party hosted API availability verified
Input
No official first-party hosted API price identified; downloadable weights are available under Apache 2.0
Output
No official first-party hosted API price identified; downloadable weights are available under Apache 2.0
Local code generation, completion, debugging, code explanation, lightweight IDE assistants, and model fine-tuning experiments
Context
131K
Reasoning
3/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API pricing; model weights are openly available
Output
No official hosted API pricing; model weights are openly available
Local code generation, completion, editing, repository-scale context, multilingual programming, and fine-tuning
Context
131K
Reasoning
6/10
Speed
7/10
Status
Available open-weight base model
Input
No official hosted API price; downloadable weights for self-hosting or third-party deployment
Output
No official hosted API price; downloadable weights for self-hosting or third-party deployment
Self-hosted coding assistance, code generation, debugging, code explanation, code translation, and long-context repository analysis
Context
131K
Reasoning
6/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight and publicly available; no verified official hosted API listing for this exact model
Input
No official hosted API price verified; downloadable weights are available under Apache 2.0
Output
No official hosted API price verified; self-hosting and third-party inference costs vary
Local bilingual image understanding, visual question answering, OCR-oriented image analysis, and image-to-text applications
Context
4K
Reasoning
5/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Open-weight multimodal model family; current accessibility and active maintenance are not clearly documented by 01.AI
Local bilingual image understanding, visual question answering, OCR-assisted extraction, image summarization, and lightweight multimodal experimentation
Context
4K
Reasoning
4/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Available as an open-weight self-hosted model; no current official hosted inference provider is listed on its model page
Self-hosted bilingual image understanding, visual question answering, image text recognition, and research applications with substantial GPU capacity
Context
4K
Reasoning
6/10
Speed
3/10
Capabilities
Image input
Multimodal input
Status
Available as open-weight downloadable model; no current first-party hosted API availability verified
Function calling, tool selection, agent orchestration, and structured workflow automation
Context
33K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Documented by 01.AI; current live availability not independently verified
Input
$3 per 1 million tokens
Output
$3 per 1 million tokens
Low-cost short-context chat, text generation, summarization, classification, and general language applications
Context
4K
Reasoning
6/10
Speed
8/10
Status
Documented in the 01.AI API catalog; current operational availability should be verified through the provider
Input
$0.19 per 1 million tokens
Output
$0.19 per 1 million tokens
Long-context text generation, open-weight research, experimentation, and private or self-hosted deployment
Context
256K
Reasoning
6/10
Speed
8/10
Status
Legacy open-weight model; downloadable and accessible through AI21 Labs' official Hugging Face repository
Long-document analysis, grounded generation, enterprise RAG, document summarization, information extraction, private deployment, and multilingual text workflows.
Context
256K
Reasoning
7/10
Speed
8/10
Status
Current and available; the API endpoint jambalarge-1.7 points to the dated snapshot jambalarge-1.7-2025-07.
Input
$2 per 1M input tokens
Output
$8 per 1M output tokens
Long-context document analysis, RAG, structured generation, function calling, multilingual enterprise assistants, and private deployment
Context
256K
Reasoning
5/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Legacy; superseded by newer Jamba releases, including AI21-Jamba-Mini-1.7. Public model weights remain available.
Long-context document analysis, retrieval-augmented generation, enterprise assistants, structured text generation, multilingual workflows, and self-hosted deployments
Context
262K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Legacy and superseded by newer Jamba Large releases; downloadable open-weight checkpoint remains available
Long-context retrieval-augmented generation, enterprise document analysis, grounded question answering, structured extraction, classification, and private deployment
Context
256K
Reasoning
6/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available; older Jamba 1.6 generation with Jamba 1.7 available as a newer successor
Input
$2 per 1 million input tokens
Output
$8 per 1 million output tokens
Long-context RAG, grounded question answering, enterprise document processing, classification, structured text generation, function calling and privacy-sensitive private deployments.
Context
256K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Legacy/open-weight model; downloadable from AI21's official Hugging Face repository and available for self-managed deployment, but not featured among AI21's current primary Jamba models as of September 25, 2026.
Long-document analysis, enterprise RAG, grounded question answering, structured text generation, private deployment, and cost-sensitive text workflows
Context
256K
Reasoning
5/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available as an AI21 open-weight model; hosted API availability is not currently verified
Input
$0.20 per 1 million input tokens when offered through AI21-hosted inference
Output
$0.40 per 1 million output tokens when offered through AI21-hosted inference
Long-context RAG, grounded enterprise question answering, document extraction, local inference, on-device assistants, and lightweight agent workflows
Context
262K
Reasoning
5/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current and available; open-weight release
Long-context enterprise question answering, grounded generation, document analysis, instruction-heavy workflows, RAG systems and self-hosted deployments
Context
256K
Reasoning
6/10
Speed
8/10
Status
Current open-weight model; available through Hugging Face and AI21 Studio
Local reasoning, long-context document analysis, private RAG, extraction, coding assistance, and lightweight agent controllers
Context
256K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current open-weight model; available for download and local inference
Input
No official AI21 hosted API price; self-hosted/open-weight model
Output
No official AI21 hosted API price; self-hosted/open-weight model
Aleph-Alpha-GermanWeb Quality Classifier
Fast German-language document quality classification and large-scale web-data filtering
Reasoning
1/10
Speed
9/10
Status
Available; open-weight research model; not deployed by a Hugging Face Inference Provider
Fast local classification of German documents for grammar-oriented corpus filtering and data curation
Status
Available as an open-weight model repository; not deployed by Hugging Face Inference Providers
GermanWeb Grammar Classifier
German grammar-related text-quality filtering, dataset curation, and binary classification of web text
Context
512
Reasoning
1/10
Speed
8/10
Status
Available as an open-weight Hugging Face model
GermanWeb Quality Classifier
German web-document quality scoring, dataset filtering, ranking, and pretraining-data curation
Context
512
Reasoning
1/10
Speed
8/10
Status
Available open-weight model
Input
No official hosted API price published
Output
No official hosted API price published
Research on tokenizer-free language modeling, English-German text generation, multilingual NLP, open-weight deployment, and custom model adaptation
Context
262K
Reasoning
5/10
Speed
4/10
Status
Available as open-weight research software through Hugging Face; not deployed by an inference provider
English and German text generation, instruction-following research, tokenizer-free language-model research, and self-hosted non-commercial experimentation
Context
262K
Reasoning
5/10
Speed
4/10
Status
Available as an open-weight research model
English and German instruction following, tokenizer-free language-model research, text-compression experiments, and self-hosted open-weight deployments
Context
98K
Reasoning
6/10
Speed
5/10
Status
Available open-weight research release
Historical multilingual text completion, language-model research, and semantic representation workflows
Reasoning
4/10
Speed
6/10
Status
Legacy; current public availability is not clearly documented
Legacy multilingual text generation, conversational applications, and explainability-oriented workflows using Aleph Alpha's first Luminous generation
Reasoning
4/10
Speed
6/10
Status
Legacy; current public availability not verified
Semantic search, information retrieval, query-document matching, clustering, classification, similarity scoring, and text feature extraction
Reasoning
2/10
Speed
7/10
Status
Legacy; current API availability unverified
Historical multilingual text-completion research and general language-task experiments
Reasoning
4/10
Speed
5/10
Status
Historical or legacy; current availability not verified
Steerable multilingual text generation, summarization, classification, question answering, and explainability-oriented enterprise workflows
Reasoning
4/10
Speed
6/10
Status
Available; first-generation Luminous control model
Input
$56.25 per 1 million tokens
Output
$61.88 per 1 million tokens
Historical multilingual text completion, language understanding research, and compatibility work involving Aleph Alpha’s original Luminous API
Reasoning
5/10
Speed
3/10
Status
Legacy generation; historical API and Playground availability documented, current public availability unverified
Zero-shot multilingual text generation, instruction following, classification, conversational prototypes, and explainability-oriented enterprise workflows.
Context
2K
Reasoning
6/10
Speed
3/10
Status
Legacy model; documented in Aleph Alpha SDK materials, with current endpoint availability dependent on provider access and deployment status.
Input
$218.75 per 1 million tokens
Output
$240.63 per 1 million tokens
Vision-language research, image captioning, visual question answering, and experiments with adapter-based multimodal fine-tuning
Reasoning
3/10
Speed
3/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Legacy research/demo model; publicly released checkpoint and source code remain available, but it is not documented as a current hosted commercial model
Input
No official hosted API pricing documented
Output
No official hosted API pricing documented
Multilingual information retrieval, semantic search, reranking, clustering, and instruction-guided text embeddings
Context
2K
Reasoning
1/10
Speed
4/10
Status
Available as downloadable open-weight model; customer and on-premises deployment options documented
Compact vector representations for semantic search, information retrieval, reranking, clustering, and similarity-based classification
Context
2K
Reasoning
1/10
Speed
6/10
Status
Available open-weight checkpoint; no hosted inference deployment or provider API pricing was verified
Concise multilingual generation, summarization, extraction, domain-specific text workflows, and self-hosted deployment
Context
8K
Reasoning
4/10
Speed
7/10
Status
Available as an open-weight model; also referenced in Aleph Alpha API tooling
Multilingual text generation, classification, summarization, question answering, engineering and automotive applications, and safety-conscious research deployments
Context
8K
Reasoning
4/10
Speed
7/10
Status
Available open-weight release; safety-aligned variant
Input
No public per-token price for the downloadable model; commercial hosted or on-premise access is available by agreement
Output
No public per-token price for the downloadable model; commercial hosted or on-premise access is available by agreement
English and German instruction following, multilingual research, tokenizer-free language-model experimentation, and self-hosted deployment
Context
262K
Reasoning
6/10
Speed
5/10
Status
Available as downloadable open-weight research software
Input
No official hosted API pricing; self-hosted deployment costs apply
Output
No official hosted API pricing; self-hosted deployment costs apply
English and German instruction following, multilingual text generation, German-language applications, and research into tokenizer-free language models
Context
164K
Reasoning
5/10
Speed
5/10
Status
Current open-weight research release; downloadable from Hugging Face
Research on tokenizer-free language modeling, English and German text generation, long-context experiments, and self-hosted non-commercial applications
Context
33K
Reasoning
5/10
Speed
5/10
Status
Current publicly downloadable open-weight research model
Local image understanding, visual question answering, captioning, document and chart analysis, counting, pointing, and multimodal research
Context
4K
Reasoning
6/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available as an open-weight downloadable checkpoint
Input
No official hosted API pricing; self-hosted weights
Output
No official hosted API pricing; self-hosted weights
Local image-and-text understanding, visual question answering, document and chart analysis, image captioning, and multimodal research
Context
4K
Reasoning
6/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight checkpoint; preview release
Input
No official hosted API pricing; downloadable weights for self-hosted deployment
Output
No official hosted API pricing; deployment cost depends on self-hosted or third-party infrastructure
Self-hosted image understanding, visual question answering, captioning, image grounding, pointing, counting, and multimodal research.
Context
4K
Reasoning
7/10
Speed
2/10
Capabilities
Image input
Multimodal input
Fine-tuning
Streaming
Status
Available open-weight legacy research model; newer Molmo 2 models are Ai2's current successor family.
Local image understanding, visual question answering, image captioning, document and chart analysis, counting, and experimentation with open multimodal models
Context
4K
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Available open-weight preview checkpoint
Efficient local image and video understanding, visual grounding, pointing, captioning, counting, tracking, and multimodal research
Reasoning
6/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Status
Current open-weight model
Input
No official hosted API pricing; self-hosted model weights
Output
No official hosted API pricing; self-hosted model weights
Open multimodal research, video question answering, visual grounding, pointing, counting, captioning, object tracking, document understanding, and robotics perception
Context
37K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current open-weight model; downloadable checkpoint
Open multimodal research, local image and video understanding, visual grounding, pointing, counting, captioning, and tracking
Context
66K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current open-weight model
Input
Not applicable; open-weight model with no official hosted API token pricing found
Output
Not applicable; open-weight model with no official hosted API token pricing found
Open research and downstream fine-tuning for vision-guided robotic manipulation, spatial reasoning, trajectory planning, and robot action prediction
Context
4K
Reasoning
7/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight research model; superseded by MolmoAct2 as Ai2's newer MolmoAct generation
Robotic manipulation research, action reasoning, downstream mid-training, and reproducing zero-shot SimplerEnv experiments
Reasoning
7/10
Speed
4/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight preview checkpoint
Input
No official hosted API pricing; self-hosted checkpoint
Output
No official hosted API pricing; self-hosted checkpoint
Open robotics research, visual action reasoning, robot-manipulation experiments, and fine-tuning on custom robot datasets
Reasoning
7/10
Speed
4/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight preview checkpoint; intended for fine-tuning and downstream post-training
Open robotics research, embodied visual reasoning, robot-policy fine-tuning, and manipulation tasks on supported or closely related hardware
Context
16K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight foundation checkpoint
Depth-aware robot manipulation research, embodied reasoning, and fine-tuning vision-language-action policies for target robot embodiments.
Reasoning
7/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight foundation checkpoint
Language-guided 3D trajectory forecasting, robotics planning research, and motion-conditioned video generation
Reasoning
2/10
Speed
4/10
Capabilities
Image input
Video input
Multimodal input
Status
Documented research variant; no separately released public checkpoint identified in the current first-party model collection
Open research and applications requiring image or video grounding, visual pointing, object localization, counting, tracking, spatial reasoning, and multimodal analysis.
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Status
Current open-weight model
Input
No official hosted API pricing; downloadable checkpoint
Output
No official hosted API pricing; downloadable checkpoint
GUI screenshot grounding, screen-element localization, computer-use perception, and research on pointing models
Context
37K
Reasoning
4/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight research model
Video object pointing, temporal grounding, object counting, tracking, and research on language-guided video understanding
Context
35K
Reasoning
4/10
Speed
5/10
Capabilities
Video input
Multimodal input
Status
Current open-weight research model
Open research, reproducible language-model experiments, local deployment, English text generation, and fine-tuning
Context
4K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model; not formally deprecated or retired
Local inference, open-model research, benchmarking, continued pretraining, and task-specific fine-tuning.
Context
4K
Reasoning
5/10
Speed
7/10
Status
Available as a downloadable open-weight model; no official first-party hosted API identified.
Input
No official Ai2 hosted API price; self-hosted model weights are downloadable.
Output
No official Ai2 hosted API price; self-hosted model weights are downloadable.
Fully open language-model research, local inference, reproducible training experiments, evaluation, and downstream fine-tuning
Context
4K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Current downloadable open-weight model; no first-party hosted API deployment verified
Input
No official hosted API price; downloadable weights are available
Output
No official hosted API price; downloadable weights are available
Continued pretraining, supervised fine-tuning, reinforcement-learning research, language-model research, custom deployment, and applications requiring a transparent open training pipeline.
Context
66K
Reasoning
5/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Current and downloadable open-weight model
Open, self-hosted chat assistants, instruction following, local coding help, tool-calling agents, and model research
Context
66K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; available for download and self-hosted inference
Open research, mathematical reasoning, code generation, logic, long-context experimentation, and local self-hosted inference
Context
66K
Reasoning
9/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Active open-weight model
Open-weight research, continued pretraining, fine-tuning, programming, mathematics, reading comprehension, and long-context language-model experiments
Context
66K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Current; open-weight and downloadable
Input
No official hosted API price; downloadable weights
Output
No official hosted API price; downloadable weights
Open-weight chat, tool-using assistants, multi-turn dialogue, synthetic data generation, self-hosting, and model research
Context
66K
Reasoning
7/10
Speed
5/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available open-weight checkpoint; superseded by Olmo 3.1 32B Instruct for newer instruction-tuned deployments
Input
No official first-party hosted API price; downloadable open weights
Output
No official first-party hosted API price; downloadable open weights
Open-model reasoning research, mathematics, coding, long-context analysis, and self-hosted deployment
Context
66K
Reasoning
8/10
Speed
5/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight model; superseded by the newer Olmo 3.1 32B Think model
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; deployment and inference costs are user-managed
Math-focused RLVR research, reinforcement-learning experiments, verifiable reward design, open-model evaluation, and further fine-tuning
Context
66K
Reasoning
7/10
Speed
7/10
Status
Current open-weight research checkpoint
Mathematical reasoning, coding, complex multi-step tasks, open-model research, local deployment, and reinforcement-learning research
Context
66K
Reasoning
8/10
Speed
5/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model
Open coding-model research, reinforcement learning from verifiable rewards, code-generation experiments, automated evaluation, and fine-tuning
Context
66K
Reasoning
5/10
Speed
7/10
Status
Current, open-weight, experimental RL-Zero coding checkpoint
Self-hosted chat assistants, instruction following, tool-use agents, synthetic data, and open-model research
Context
66K
Reasoning
7/10
Speed
4/10
Capabilities
Tool use
Fine-tuning
Status
Current; open-weight and downloadable
English audio transcription, captions, meetings, lectures, podcasts, accessibility applications, and open ASR research
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Streaming
Status
Current open-weight model
Open English speech-to-text transcription, captioning, meetings, lectures, calls, podcasts, and local ASR research
Reasoning
1/10
Speed
6/10
Capabilities
Audio input
Streaming
Status
Available open-weight model
Input
No official hosted API pricing; released for open-weight deployment
Output
No official hosted API pricing; released for open-weight deployment
Open, self-hosted English transcription for meetings, lectures, calls, podcasts, accessibility, and speech-recognition research
Reasoning
1/10
Speed
5/10
Capabilities
Audio input
Streaming
Status
Current open-weight model; downloadable and usable for local inference
Input
No official hosted API pricing; self-hosted open weights
Output
No official hosted API pricing; self-hosted open weights
English short- and long-form speech transcription, meeting and call transcription, lecture captioning, podcast processing, broadcast transcription, and speech analytics
Reasoning
1/10
Speed
6/10
Status
Current open-weight model
Input
No official hosted API price; self-hosted checkpoint
Output
No official hosted API price; self-hosted checkpoint
Open English speech transcription, meeting and podcast transcription, captioning, timestamped audio indexing, and ASR research
Reasoning
1/10
Speed
7/10
Status
Current open-weight model; available for download and self-managed inference
Input
No official hosted API pricing; open weights available for self-hosted use
Output
No official hosted API pricing; output is transcribed text generated during local inference
Efficient self-hosted English transcription, speech research, edge-oriented experiments, and applications where a small open ASR checkpoint is preferred.
Reasoning
1/10
Speed
8/10
Status
Available open-weight checkpoint
Input
No official hosted API price; self-hosted checkpoint
Output
No official hosted API price; self-hosted checkpoint
Open-model research, efficient local text generation, language-model evaluation, and fine-tuning
Context
4K
Reasoning
4/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight release
Efficient satellite-image and Earth-observation embeddings, remote-sensing research, and downstream classification or segmentation
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current; downloadable open-weight research model
Satellite-image and Earth-observation embeddings, remote-sensing representation learning, geospatial classification, segmentation, and downstream fine-tuning
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight model
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; produces feature representations and embeddings
Efficient Earth observation embeddings, satellite image and time-series representation learning, geospatial classification, segmentation, and large-scale remote-sensing inference
Reasoning
2/10
Speed
9/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight model
Satellite-image embeddings, remote-sensing representation learning, geospatial segmentation, land-cover analysis, and Earth observation research
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight Earth observation foundation model
Input
No official public hosted API price; downloadable weights are available
Output
No official public hosted API price; model output is primarily embeddings and task representations
Open-weight language-model research, long-context text generation, local deployment, continued pretraining, and fine-tuning
Context
66K
Reasoning
6/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model
Self-hosted instruction following, open post-training research, mathematics, coding, evaluation and domain adaptation
Context
131K
Reasoning
7/10
Speed
7/10
Status
Available for download; superseded by Llama-3.1-Tulu-3.1-8B
Self-hosted instruction following, reasoning, mathematics, coding, open-model research, and reproducible post-training experiments
Context
131K
Reasoning
8/10
Speed
3/10
Status
Available open-weight model
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; downloadable weights
Large-scale research, instruction-following evaluation, open-weight post-training research, mathematical reasoning, coding benchmarks, and self-hosted experimentation.
Context
8K
Reasoning
8/10
Speed
2/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight research and educational model; no hosted inference provider deployment is currently listed on its Hugging Face model page.
Input
No official first-party hosted API price; self-hosted open weights
Output
No official first-party hosted API price; self-hosted open weights
Multimodal research, vision-language experiments, image generation, visual question answering, dense computer-vision tasks, and academic benchmarking
Reasoning
4/10
Speed
3/10
Capabilities
Image input
Multimodal input
Status
Open-weight research release; publicly available inference code and checkpoints; legacy relative to Ai2's newer multimodal models
Multimodal research, image understanding and generation, audio and video understanding, spatial prediction, embodied AI and robotic-manipulation experiments, and self-hosted academic prototyping.
Reasoning
5/10
Speed
2/10
Outputs
Text
Image
Speech
Music
Actions
Capabilities
Image input
Audio input
Video input
Multimodal input
Fine-tuning
Status
Open-weight research release; publicly accessible checkpoints and source code; no official hosted inference API or commercial token pricing identified.
Input
No official hosted API pricing; self-hosted research checkpoints
Output
No official hosted API pricing; self-hosted research checkpoints
Open-vocabulary monocular 3D detection, spatial perception, robotics research, augmented reality, and lifting 2D prompts into metric 3D boxes
Reasoning
2/10
Speed
2/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight research model
Input
No official hosted API pricing; downloadable checkpoint
Output
No official hosted API pricing; downloadable checkpoint
Reasoning
0/10
Speed
0/10
Enterprise image generation and editing, product visualization, advertising and marketing assets, image variations, background removal, virtual try-on, and brand or subject-consistent visual content.
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Legacy; currently accessible as of 2026-09-25; scheduled for end of life on 2026-09-30
Low-cost multimodal document analysis, image and video understanding, visual question answering, summarization, RAG, and tool-enabled agents.
Context
300K
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Input
$0.06 per 1 million input tokens
Output
$0.24 per 1 million output tokens
High-volume, low-latency text classification, summarization, translation, extraction, routing, FAQs and narrowly defined fine-tuned tasks
Context
128K
Reasoning
4/10
Speed
10/10
Capabilities
Tool use
Fine-tuning
Streaming
Input
$0.035 per 1 million input tokens
Output
$0.14 per 1 million output tokens
Long-context multimodal analysis, enterprise document workflows, complex tool calling, agentic orchestration, codebase analysis, and teacher-model distillation before retirement.
Context
1M
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Enterprise multimodal applications, document analysis, visual question answering, video understanding, long-context summarization, RAG, and tool-using assistants
Context
300K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Input
$0.80 per 1 million input tokens
Output
$3.20 per 1 million output tokens
Short-form advertising, marketing concepts, product visualization, storyboards, social video drafts, and image-guided cinematic clips.
Capabilities
Image input
Multimodal input
Status
Active through the current Nova Reel 1.1 workflow; the original Nova Reel v1.0 model is legacy and scheduled for end of life on 2026-09-30.
Input
Not token-priced; video generation is priced per generated video second.
Output
$0.08 per generated video second, subject to AWS region, pricing-tier, and current Bedrock pricing conditions.
Real-time voice assistants, customer-service automation, interactive education, language learning, and speech-enabled enterprise workflows
Context
300K
Reasoning
5/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Retired; legacy model with official end-of-life date of September 14, 2026
High-volume multimodal applications, document and video analysis, customer service, business automation, software engineering, long-context workflows, and cost-sensitive AI agents.
Context
1M
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Active; generally available through Amazon Bedrock. AWS states EOL is no sooner than 2026-12-02.
Input
$0.30 per 1 million input tokens on the global standard rate; regional and service-tier prices may vary.
Output
$2.50 per 1 million output tokens on the global standard rate; regional and service-tier prices may vary.
Real-time voice assistants, customer-service automation, telephony, interactive learning, multilingual conversations, and tool-enabled speech agents.
Context
1M
Reasoning
6/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Active; EOL no sooner than 2026-12-02
Input
$0.003 per 1,000 speech-input units; text-input pricing may apply separately and should be verified on the current Amazon Bedrock pricing page.
Output
$0.012 per 1,000 speech-output units; text-output pricing may apply separately and should be verified on the current Amazon Bedrock pricing page.
Browser automation, visual UI navigation, repetitive web workflows, agentic QA, tool-oriented tasks, and human-supervised enterprise processes
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Status
Generally available
Input
$4.75 per agent hour; token-level input pricing is not published for this model
Output
$4.75 per agent hour; token-level output pricing is not published for this model
Amazon Nova Multimodal Embeddings
Cross-modal semantic search, multimodal RAG, digital asset discovery, recommendations, classification, and clustering
Context
8K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Active; generally available through Amazon Bedrock
Input
Modality-dependent pricing; AWS charges based on processed input and processing mode. Current pricing should be checked on the Amazon Bedrock pricing page.
Output
Not applicable as output-token pricing; the model returns embeddings and pricing is based primarily on input modality and processing mode.
Multimodal search, text-to-image retrieval, image similarity, visual recommendations, personalization, and image-text matching.
Context
256
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Input
Text: $0.0008 per 1,000 input tokens; images: $0.00006 per input image. Pricing may vary by AWS Region and current Bedrock pricing terms.
Output
No separate output-token price; the model returns embedding vectors.
Text-to-image generation, image editing, reference-guided composition, background removal, color-controlled visuals, image variations, and subject-consistent branded content.
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Retired; AWS lists the model as legacy with an end-of-life date of 2026-06-30.
Input
Not token-priced; image-generation pricing applies. AWS pricing examples list $0.01 per 1,024×1,024 standard-quality image.
Output
$0.01 per 1,024×1,024 standard-quality image in the AWS pricing example; smaller images and premium quality use different pricing tiers.
Semantic search, vector indexing, retrieval-augmented generation, personalization, clustering, classification, and recommendation pipelines.
Context
8K
Reasoning
1/10
Speed
8/10
Status
Available; older G1/V1 text-embedding generation
Input
$0.10 per 1 million input tokens
Cost-efficient text embeddings for semantic search, RAG, document retrieval, classification, clustering, reranking, and recommendations
Context
8K
Reasoning
1/10
Speed
8/10
Input
$0.02 per 1 million input tokens
Semantic search, vector retrieval, recommendation, semantic matching, knowledge bases, and retrieval-augmented generation
Status
Current and accessible through Baidu Qianfan and AI Studio embedding APIs
Input
Â¥0.0005 per 1,000 input tokens
Output
Â¥0.0005 per 1,000 completion tokens in current model-list metadata; the model practically returns vector embeddings rather than generated completion text
Multimodal understanding, long-context Chinese and English applications, complex reasoning, coding, tool-enabled agents, and enterprise workloads
Context
249K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current and accessible through Baidu Qianfan as of September 25, 2026
Input
¥0.006 per 1K input tokens for requests up to 32K tokens; ¥0.010 per 1K input tokens above 32K
Output
¥0.024 per 1K output tokens for requests up to 32K tokens; ¥0.040 per 1K output tokens above 32K
Realistic text-to-image generation, reference-grounded visual creation, commercial-style imagery, and applications needing reduced generative-artificiality.
Reasoning
2/10
Speed
8/10
Status
Available; listed in Baidu AI Cloud's inference-service model updates and supported for offline batch inference in Qianfan ModelBuilder documentation.
Object removal, masked image repainting, image variation, and batch image-editing workflows.
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Available with access dependent on Baidu Qianfan service configuration; listed for Qianfan batch inference. Related AI作画-iRAG版 API sales stopped on 2026-04-30, but that product notice does not explicitly state that the Qianfan ERNIE-iRAG-Edit endpoi
Lightweight local text generation, Chinese and English language experimentation, compact conversational systems, and domain adaptation
Context
131K
Reasoning
3/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model
Fast, low-cost Chinese text generation, long-context applications, enterprise agents, content creation, reasoning, and code assistance.
Context
138K
Reasoning
7/10
Speed
9/10
Capabilities
Tool use
Web search
Status
Current and accessible through Baidu Qianfan; canonical API endpoint is ernie-4.5-turbo-128k.
Input
¥0.0008 per 1,000 input tokens; web-search augmentation is priced at ¥0.004 per 1,000 tokens where applicable.
Output
Â¥0.0032 per 1,000 output tokens.
Cost-efficient multimodal understanding, image and video analysis, OCR, document comprehension, translation, visual question answering, and code-related tasks
Context
128K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and available
Input
Â¥0.003 per 1,000 tokens
Output
Â¥0.009 per 1,000 tokens
Local or self-hosted multimodal applications, visual question answering, document and chart understanding, image analysis, video understanding, and efficient vision-language inference
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Streaming
Status
Current open-weight model; publicly released under the Apache 2.0 license
Agentic workflows, web-search-assisted tasks, reasoning, Chinese-language knowledge work, creative writing, and general-purpose assistant applications.
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Web search
Status
Current; officially released and accessible through Baidu's ERNIE website and AI Studio playground. Public Qianfan API availability for an exact ERNIE 5.1 endpoint was not verified.
Code completion, code generation, unit-test generation, code optimization, code explanation, and fine-tuning for software-development workflows.
Context
128K
Reasoning
4/10
Speed
6/10
Status
Retired; Qianfan ModelBuilder listed August 14, 2025 as the retirement date.
Chinese-language reasoning, long-form analysis, complex calculations, literary and document writing, agent workflows, and function calling
Context
33K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Web search
Streaming
Status
Ready; currently accessible through Baidu Qianfan as ernie-x1-turbo-32k
Deep reasoning, Chinese and English question answering, mathematics, coding, factual responses, tool calling, web-grounded applications, and agent workflows
Context
66K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Web search
Streaming
Status
Current preview model
Input
Â¥0.001 per 1K input tokens
Output
Â¥0.004 per 1K output tokens
Low-cost image-to-video generation, short social clips, product animation, marketing assets, and turning still images into dynamic scenes
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through Baidu Qianfan API
Input
CNY 1.00 per 5-second video
Chinese image-to-video generation, audiovisual storytelling, marketing videos, multi-person dialogue, synchronized speech, sound effects, and cinematic short-form content
Capabilities
Image input
Multimodal input
Status
Current model family; concrete Qianfan variants include Turbo, Lite, Pro, Turbo-I2V-Audio, and Turbo-I2V-Effect
Low-cost text-to-image generation, marketing visuals, creative assets, illustrations, and rapid image prototyping
Reasoning
1/10
Speed
8/10
Status
Current and available
Input
Â¥0.05 per image at 1024x1024
Output
Â¥0.05 per generated image at 1024x1024
Multilingual OCR and structured parsing of complex documents, including tables, formulas, charts, seals, scanned pages, warped documents, and screen photographs
Context
131K
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Available; superseded by PaddleOCR-VL-1.6
Document parsing, OCR, layout analysis, table extraction, and structured document understanding
Capabilities
Image input
Multimodal input
Fast enterprise question answering, summarization, workflow nodes, agent response generation, and text processing with a 32K context window
Context
33K
Reasoning
5/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Available through Baidu’s documented Wenxin Workshop API; not found in the current Qianfan V2 model-list documentation, so new integrations should verify legacy endpoint compatibility.
Long-context Agent planning, task decomposition, component selection, and function-calling workflows.
Context
128K
Reasoning
6/10
Speed
8/10
Status
Current platform-listed planning model; standalone lifecycle status is not separately published.
Enterprise agent workflows, intent recognition, instruction routing, and tool-calling tasks
Context
33K
Reasoning
4/10
Speed
8/10
Status
Legacy; current public availability not verified
Fast enterprise question answering, lightweight agent planning, component selection, and streaming text applications
Context
8K
Reasoning
4/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current and listed as available in Baidu Qianfan documentation; fast agent-oriented model
Long-horizon, high-precision dexterous robot manipulation and real-world VLA policy specialization
Reasoning
7/10
Speed
4/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current research model/framework; public API availability not documented
Protein and biomolecular complex structure prediction, computational biology research, molecular design workflows, and self-hosted scientific inference
Reasoning
8/10
Speed
5/10
Status
Active open-source biomolecular structure prediction project with multiple model variants, including Protenix-v1 and Protenix-v2
Multimodal agent workflows, search and information retrieval, coding agents, GUI interaction, image and video understanding, complex instruction following, and long-context business tasks.
Context
256K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Beta; accessible through BytePlus ModelArk as seed-1-8-251228. The separate LAS multimodal deep-thinking operator using Seed1.8 ended service on 2026-09-20.
Input
USD 0.25 per million tokens for prompts up to 128K tokens; USD 0.50 per million tokens for prompts over 128K and up to 256K; cached input USD 0.05 per million tokens.
Output
USD 2.00 per million tokens for prompts up to 128K tokens; USD 4.00 per million tokens for prompts over 128K and up to 256K. Batch output pricing is USD 1.00 or USD 2.00 per million tokens by prompt-length tier.
Self-hosted general language modeling, long-context research, reasoning experiments, coding assistance, summarization, and foundation-model fine-tuning
Context
524K
Reasoning
7/10
Speed
5/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API price; downloadable weights are available under Apache-2.0
Output
No official hosted API price; deployment cost depends on user infrastructure or third-party hosting
Self-hosted long-context reasoning, coding, agentic tool use, research, question answering, summarization, and general text generation
Context
524K
Reasoning
8/10
Speed
5/10
Capabilities
Tool use
Streaming
Status
Current open-weight model; downloadable under the Apache-2.0 license
Research, custom post-training, long-context processing, coding experiments, and self-hosted foundation-model deployment
Context
512K
Reasoning
7/10
Speed
4/10
Capabilities
Tool use
Fine-tuning
Status
Current open-weight downloadable foundation model
General-purpose Chinese and multilingual assistance, coding, reasoning, image and document understanding, and voice-interaction applications.
Context
33K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Web search
Status
Retired; the primary doubao-1-5-pro-32k-250115 deployment was scheduled to shut down on 2026-09-21 at 14:00 China Standard Time.
Visual reasoning, image and video understanding, OCR, visual grounding, GUI-agent research, gameplay analysis, and multimodal benchmark evaluation
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Retired; Volcano Engine service ended on 2026-03-31
Cross-modal semantic search, text-image retrieval, video retrieval, multimodal knowledge bases, classification, clustering, and recommendation
Context
128K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Status
Current and available through Volcano Engine
Multimodal document analysis, visual question answering, coding, mathematics, general reasoning, long-context analysis and adaptive-thinking applications
Context
256K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Deprecated; new endpoint creation stopped on 2026-09-24; existing service scheduled for automatic migration or replacement on 2026-11-24
Deep reasoning, coding, mathematics, logical analysis, document understanding, and visual reasoning over images or videos
Context
256K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Deprecated; scheduled for retirement and migration to a Seed2.0 model
Cost-conscious production applications requiring long-context multimodal understanding, document and video analysis, coding assistance, tool use, GUI automation, and structured extraction.
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; latest documented release seed-2-0-lite-260428
Input
$0.25 per 1M non-audio input tokens for prompts up to 128K; $0.50 per 1M for prompts above 128K and up to 256K. Audio input: $3.75 per 1M tokens up to 128K and $7.50 per 1M above 128K.
Output
$2.00 per 1M output tokens for prompts up to 128K; $4.00 per 1M for prompts above 128K and up to 256K.
High-concurrency inference, batch generation, classification, extraction, summarization, and cost-sensitive multimodal workloads
Context
256K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Status
Current; available through ByteDance's Volcano Engine model API
Input
$0.03 per 1M input tokens
Output
$0.31 per 1M output tokens
Complex multimodal reasoning, long-chain agent workflows, visual and video analysis, document understanding, scientific research support, coding, and enterprise automation.
Context
200K
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Active and currently accessible through Volcano Engine Ark; canonical deployment version 260215
Input
CNY 3.2 per million tokens for 0–32K input; CNY 4.8 per million tokens for 32–128K input; CNY 9.6 per million tokens for 128–256K input. Batch inference input pricing is CNY 1.6, 2.4, and 4.8 per million tokens for the same tiers.
Output
CNY 16 per million tokens for 0–32K input; CNY 24 per million tokens for 32–128K input; CNY 48 per million tokens for 128–256K input. Batch inference output pricing is CNY 8, 12, and 24 per million tokens for the same tiers.
Complex agent workflows, high-value office and research tasks, long-horizon coding, document and visual analysis, video understanding, and tool-enabled productivity automation.
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Status
Current and officially released; available through ByteDance Seed, Doubao, and Volcano Engine API channels.
Fast multimodal agents, coding assistants, document and video analysis, tool-calling workflows, structured data extraction, and cost-sensitive production applications
Context
256K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and available through Volcano Engine Ark; API model identifier doubao-seed-2-1-turbo-260628
Input
CNY 6 per 1M input tokens for input lengths up to 256K tokens
Output
CNY 30 per 1M output tokens
Long-form audiovisual storytelling, text-to-video, reference-based video generation, creative production, advertising, education, industrial simulation, and video editing.
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Current; available through ByteDance platforms including Jimeng AI, Doubao Pro, and the Seed platform. API access through BytePlus ModelArk was announced as forthcoming.
Multimodal text-to-video and reference-based video creation, cinematic short clips, video editing and extension, multi-shot storytelling, and synchronized audio-video production.
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Current official model page remains available; older generation superseded by Seedance 2.5
Full-scene audio creation, expressive voice generation, dubbing, dialogue, sound effects, ambience, advertising, games, podcasts, and multilingual audio production
Capabilities
Image input
Audio input
Multimodal input
Status
Currently available through BytePlus
High-throughput code generation research, diffusion-language-model evaluation, and code-editing experiments
Reasoning
4/10
Speed
10/10
Status
Experimental research preview
Embodied robotics research, long-horizon manipulation, bimanual control, dexterous object handling, and adapting robot policies to new objects and tasks
Reasoning
6/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current officially documented research model; no public hosted API, commercial pricing, or downloadable weights verified
Real-time audio-visual assistants, scene-aware guidance, live explanation, interactive learning, accessibility, and proactive multimodal collaboration
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; fully rolled out for large-scale audio-visual full-duplex deployment
Professional image generation and editing, high-density infographics, advertising and e-commerce creative, multilingual visual content, precise spatial edits, layer separation, and multi-reference image workflows
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current and publicly documented; available through BytePlus ModelArk and ByteDance Seed services
Input
First reference image free; additional reference images $0.003 per image
Output
$0.045 per output image for images up to 2.36 million pixels; $0.09 per output image above 2.36 million pixels
Open-weight computer-use research, GUI grounding, browser automation prototypes, screenshot-based interface interaction, and visual action-model experimentation
Context
128K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Multimodal input
Status
Available open-weight research release; superseded by UI-TARS-2 for the provider's newer UI-TARS development line
Research, local text generation, fine-tuning experiments, benchmarking, and lightweight English NLP prototypes
Context
2K
Reasoning
2/10
Speed
7/10
Status
Available open-weight research model
Open-weight language-model research, local text generation, fine-tuning experiments, scaling-law studies, and educational or reference implementations.
Context
2K
Reasoning
3/10
Speed
7/10
Status
Available as an open-weight research model
Open LLM research, scaling-law experiments, language-model evaluation, fine-tuning, and self-hosted English text generation
Context
2K
Reasoning
4/10
Speed
7/10
Status
Open-weight and publicly downloadable; research-oriented base model
Open LLM research, local text generation, reproducible training experiments, and downstream fine-tuning
Context
2K
Reasoning
2/10
Speed
4/10
Status
Open-weight, publicly downloadable; no official retirement date identified
Research, small-scale language-model experimentation, causal text generation, fine-tuning studies, and local deployment
Context
2K
Reasoning
2/10
Speed
8/10
Status
Available open-weight model; legacy research release
Compact local language-model experiments, text completion, education, benchmarking, and fine-tuning research
Context
2K
Reasoning
2/10
Speed
8/10
Status
Available as downloadable open weights; research-oriented and not instruction-tuned
Local text generation, language-model research, benchmarking, fine-tuning experiments, and studying compute-efficient scaling at small model sizes
Context
2K
Reasoning
2/10
Speed
8/10
Status
Available open-weight research checkpoint
Multi-turn document retrieval and retrieval-augmented generation pipelines
Reasoning
2/10
Speed
8/10
Status
Available as open-weight query and context encoder checkpoints
Document-grounded conversational question answering, local retrieval-augmented generation, and open-weight research deployments
Context
8K
Reasoning
5/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Available as a public open-weight checkpoint; not identified as a current Cerebras-hosted API model
Low-latency assistants, customer support, high-volume text processing, coding assistance, image understanding, and parallel subagent workloads
Context
200K
Reasoning
8/10
Speed
10/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active (latest); retirement not sooner than October 15, 2026
Input
$1 per million input tokens; cache write $1.25 per million tokens for 5 minutes or $2 per million tokens for 1 hour; cache read $0.10 per million tokens
Output
$5 per million output tokens; batch API output pricing is 50% lower
Complex reasoning, long-running autonomous agents, advanced coding, multi-stage research, document-heavy analysis, and high-value enterprise workflows
Context
1M
Reasoning
10/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active; tentative retirement not sooner than 2027-06-09
Input
$10 per million input tokens; $12.50 per million tokens for 5-minute cache writes; $20 per million tokens for 1-hour cache writes; $1 per million tokens for cache reads
Output
$50 per million output tokens; $25 per million output tokens for batch processing
Long-running agentic coding, demanding reasoning, multistep research, complex document analysis, spreadsheets, presentations, and high-stakes knowledge work
Context
1M
Reasoning
10/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$10 per million input tokens; 5-minute cache writes $12.50 per million tokens; 1-hour cache writes $20 per million tokens; cache reads $0.25 per million tokens
Output
$50 per million output tokens; batch output pricing $25 per million tokens
Advanced cybersecurity, vulnerability research, biology, healthcare, life-sciences research, and long-running technical workflows requiring extensive context and reasoning
Context
1M
Reasoning
10/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active; invite-only; preview/beta on Amazon Bedrock
Input
$10 per million input tokens; cached input: $1 per million tokens for cache reads; cache writes: $12.50 per million tokens for 5-minute TTL or $20 per million tokens for 1-hour TTL
Output
$50 per million output tokens; Batch API input and output receive a 50% discount
Vetted cybersecurity defense, advanced biology and life-sciences research, complex coding, long-running agents, technical investigation, and high-value research workflows
Context
1M
Reasoning
10/10
Speed
4/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Active; invitation-only through Project Glasswing and trusted-access programs
Input
$10 per million input tokens; 5-minute cache write $12.50 per million tokens; 1-hour cache write $20 per million tokens; cache read $0.25 per million tokens
Output
$50 per million output tokens; Batch API offers a 50% discount on input and output pricing
Defensive cybersecurity research, vulnerability discovery, attack-surface auditing, autonomous coding, long-running agents, and large-context analysis.
Context
1M
Reasoning
9/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Deprecated; deprecated June 9, 2026. Anthropic recommends migrating to Claude Mythos 5.
Complex reasoning, agentic coding, repository-scale software work, long-context analysis, research, document and spreadsheet workflows, and enterprise automation
Context
1M
Reasoning
10/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active legacy; retirement not sooner than 2027-02-05
Input
$5 per million input tokens; cache write $6.25 per million tokens for 5 minutes or $10 per million tokens for 1 hour; cache read $0.50 per million tokens
Output
$25 per million output tokens; Batch API input and output pricing discounted 50%
Complex agentic coding, code review, enterprise analysis, long-context research, document workflows, financial and legal reasoning, and multi-step tool-using applications.
Context
1M
Reasoning
10/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active legacy; Anthropic recommends migration to Claude Opus 5.5. Retirement is scheduled no sooner than 2027-07-24.
Input
$5 per million tokens; cache writes $6.25 per million tokens for 5 minutes or $10 per million tokens for 1 hour; cache reads $0.50 per million tokens; batch input receives a 50% discount.
Output
$25 per million tokens; batch output receives a 50% discount.
Long-running agentic coding, repository-scale software engineering, code review, knowledge work, multimodal document analysis, and tool-using applications
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$4 per million input tokens
Output
$20 per million output tokens
Complex software engineering, coding agents, multi-step research, computer-use workflows, enterprise analysis, visual document understanding, and high-value tool-using applications.
Context
200K
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$5 per million input tokens
Output
$25 per million output tokens
Complex reasoning, agentic coding, repository-scale software engineering, research, document analysis, computer-use workflows, and high-stakes knowledge work
Context
1M
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active legacy; migration to Claude Opus 5.5 recommended
Input
$5 per million input tokens; 5-minute cache write $6.25 per million; 1-hour cache write $10 per million; cache read $0.50 per million; Batch API $2.50 per million
Output
$25 per million output tokens; Batch API $12.50 per million
Complex reasoning, agentic coding, long-horizon workflows, enterprise knowledge work, document analysis, vision tasks, and computer-use agents
Context
1M
Reasoning
10/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active legacy model; migration to newer Opus models is recommended for new deployments
Input
$5 per million input tokens
Output
$25 per million output tokens
Complex coding, software agents, computer-use workflows, visual document analysis, research, and long-running multi-step tasks
Context
200K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Legacy; still available as of 2026-09-24; migration to Claude Sonnet 5 recommended
Input
$3 per million tokens; cache write $3.75 per million tokens for 5 minutes or $6 per million tokens for 1 hour; cache read $0.30 per million tokens
Output
$15 per million tokens; Message Batches API offers a 50% discount on input and output pricing
Agentic coding, software engineering, browser and computer-use workflows, long-context analysis, tool-driven automation, and high-volume assistants
Context
1M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current and generally available
Input
$2 per million input tokens
Output
$10 per million output tokens
Agentic coding, computer use, long-context document analysis, enterprise knowledge work, structured extraction, and high-volume applications needing strong reasoning at Sonnet-tier pricing
Context
1M
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Active legacy; Anthropic recommends migration to Claude Sonnet 5
Input
$3 per million input tokens
Output
$15 per million output tokens
Fast coding agents, long-context analysis, image-aware workflows, document creation, and tool-enabled business applications
Context
1M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$2 per million tokens
Output
$10 per million tokens
Multilingual generation, translation, summarization, customer support, global communication, and multilingual research
Context
128K
Reasoning
6/10
Speed
4/10
Input
$0.50 per 1 million tokens
Output
$1.50 per 1 million tokens
Multilingual image understanding, OCR, image captioning, visual question answering, image-based translation, visual reasoning and research deployments using open weights.
Context
16K
Reasoning
7/10
Speed
5/10
Capabilities
Image input
Multimodal input
Streaming
Multilingual audio transcription, enterprise speech archives, meeting and interview transcription, call-center audio, and low-latency ASR workflows
Reasoning
2/10
Speed
8/10
Status
Live; open-source research release
Input
No per-input token price published. API experimentation is free subject to rate limits; Model Vault uses hourly instance-based pricing.
Output
No per-output token price published. API experimentation is free subject to rate limits; Model Vault uses hourly instance-based pricing.
Arabic speech transcription, multilingual Arabic-English audio, regional dialects, code-switched speech, call-center audio and high-throughput ASR workloads.
Reasoning
1/10
Speed
8/10
Input
Free through Cohere API for experimentation subject to rate limits; Model Vault deployment is priced per hour-instance and requires contacting Cohere.
Enterprise RAG, long-context document analysis, multilingual applications, tool use, agentic workflows, financial text processing and structured text generation
Context
256K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Streaming
Input
$2.50 per 1 million tokens
Output
$10.00 per 1 million tokens
Enterprise agents, multimodal document and image analysis, multilingual workflows, reasoning-intensive automation, retrieval-augmented generation, and tool-using applications.
Context
128K
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Input
Free until applicable rate limits; production pricing is not publicly listed and may require contacting Cohere.
Output
Free until applicable rate limits; production pricing is not publicly listed and may require contacting Cohere.
Complex enterprise agents, tool use, retrieval-augmented generation, multilingual reasoning, long-context analysis, and workflow automation.
Context
256K
Reasoning
8/10
Speed
5/10
Capabilities
Tool use
Web search
Streaming
Input
Free until applicable API rate limits; private Model Vault deployment is billed per instance-hour rather than per token. Standard Vault pricing lists $48/hour for L and $57.50/hour for XL performance tiers.
Output
Free until applicable API rate limits on Cohere-hosted trial and production access; Model Vault pricing is instance-hour based rather than output-token based.
High-quality multilingual text translation, enterprise document translation, and privacy-sensitive translation workflows.
Context
8K
Reasoning
4/10
Speed
7/10
Capabilities
Tool use
Streaming
Input
Free until applicable rate limits are reached; production access requires contacting Cohere sales.
Output
Free until applicable rate limits are reached; production access requires contacting Cohere sales.
Enterprise document intelligence, OCR, chart and table analysis, visual question answering, and multilingual image understanding
Context
128K
Reasoning
6/10
Speed
7/10
Capabilities
Image input
Multimodal input
Streaming
Input
Free until applicable rate limits; production access requires contacting Cohere
Output
Free until applicable rate limits; production access requires contacting Cohere
Low-cost enterprise RAG, multilingual document workflows, long-context chat, structured extraction, and tool-using agents
Context
128K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Input
$0.15 per 1M input tokens
Output
$0.60 per 1M output tokens
Cost-sensitive RAG, enterprise chat, tool use, coding assistance, and fast multi-step agents
Context
128K
Reasoning
7/10
Speed
9/10
Capabilities
Tool use
Web search
Streaming
Input
$0.0375 per 1 million tokens
Output
$0.15 per 1 million tokens
Complex enterprise RAG, long-context document analysis, multilingual assistants, citations, structured data tasks and multi-step tool-use agents.
Context
128K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Streaming
Input
$2.50 per 1 million tokens
Output
$10 per 1 million tokens
Multilingual semantic search, multimodal RAG, PDF and document retrieval, image-to-text retrieval, classification, clustering, and enterprise vector indexing.
Context
128K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Generally available
Input
$0.12 per 1 million text input tokens; $0.47 per 1 million image tokens
Fast, storage-efficient English semantic search, retrieval, classification, clustering, and large-scale embedding workloads
Capabilities
Image input
Multimodal input
Status
Current and available
English semantic search, retrieval-augmented generation, vector indexing, classification, clustering, and similarity matching
Capabilities
Image input
Multimodal input
Input
$0.10 per 1 million input tokens
Output
Not applicable; the model returns embeddings rather than billed generated text
Fast multilingual semantic search, cross-lingual retrieval, RAG, clustering, classification features, and compact vector indexes
Capabilities
Image input
Multimodal input
Multilingual semantic search, cross-lingual retrieval, RAG indexing, classification, clustering, and image-text similarity
Context
512
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Input
$0.10 per 1M input tokens
Agentic software engineering, repository-level code changes, terminal-based coding agents, code review, local inference, and private deployment.
Context
256K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Input
Free until applicable Cohere API rate limits are reached; self-hosted deployment has infrastructure costs rather than Cohere per-token pricing.
Output
Free until applicable Cohere API rate limits are reached; self-hosted deployment has infrastructure costs rather than Cohere per-token pricing.
Enterprise machine translation, multilingual documentation, localization, internal communications, safety procedures, and private or self-hosted translation workflows
Context
16K
Reasoning
6/10
Speed
6/10
Capabilities
Tool use
Streaming
Input
Cohere API: free until applicable rate limits are reached; Standard Model Vault: $17.50 per instance-hour for XL deployment
Output
No separate token-based output price published; Cohere API is free until applicable rate limits are reached
Compact multilingual image understanding, OCR, documents, charts, visual question answering, and specialized fine-tuning
Context
128K
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight release
High-volume enterprise document parsing, table and form extraction, search indexing, RAG ingestion, and document context for AI agents
Context
8K
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Live; generally available
Input
$1.50 per 1,000 pages through the Cohere API
Multilingual semantic reranking, enterprise search, hybrid retrieval, and RAG pipelines
Context
4K
Reasoning
3/10
Speed
8/10
English semantic reranking for enterprise search, hybrid retrieval, RAG pipelines, FAQs, knowledge bases, documents, code retrieval, and semi-structured records
Context
4K
Reasoning
2/10
Speed
8/10
Status
Active and currently listed by Cohere; older Rerank 3.0 English model with newer Rerank alternatives available
Input
$2.00 per 1,000 search units; one search unit covers one query with up to 100 documents, subject to chunking
Multilingual semantic reranking for enterprise search, hybrid retrieval, cross-language search, and RAG pipelines
Context
4K
Reasoning
5/10
Speed
7/10
Status
Current older-generation model; superseded by newer Rerank model generations
Low-latency multilingual search reranking, high-throughput retrieval, enterprise search, hybrid search and RAG pipelines
Context
33K
Reasoning
1/10
Speed
9/10
Status
Current and available
Input
Usage-based search-unit pricing; one search unit is one query with up to 100 documents to be ranked
High-quality multilingual reranking for enterprise search, RAG pipelines, semantic retrieval, and semi-structured document ranking.
Context
33K
Reasoning
4/10
Speed
7/10
Status
Generally available
Input
$2.50 per 1,000 search units; one search unit is one query against up to 100 documents. Dedicated Model Vault pricing is separate.
Output
Not applicable; the endpoint returns relevance scores and ranking results rather than generated tokens.
Multilingual translation, conversation, summarization, and text generation focused on African and West Asian languages; local and edge deployment
Context
8K
Reasoning
3/10
Speed
8/10
South Asian multilingual conversation, translation, target-language generation, local inference, and edge or on-device applications
Context
8K
Reasoning
3/10
Speed
8/10
Multilingual translation, cross-lingual text generation, localized assistants, education, research, and efficient local or edge deployment
Context
8K
Reasoning
3/10
Speed
8/10
Efficient multilingual translation, text generation, localization, language learning, and local or edge deployment for European and Asia-Pacific languages
Context
8K
Reasoning
3/10
Speed
8/10
Status
Live; available through the Cohere Chat API and as an open-weight model
Self-hosted English text completion, code completion, model research, and fine-tuning experiments using an open-weight mixture-of-experts model.
Context
33K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Databricks Foundation Model APIs and Foundation Model Fine-tuning; open weights remain available for self-hosted use.
Self-hosted general English-language instruction following, coding assistance, text generation, and domain-specific fine-tuning
Context
33K
Reasoning
7/10
Speed
5/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight model; Databricks Foundation Model API serving retired
Input
10.714 DBU per 1 million input tokens when historically available through Databricks pay-per-token serving; endpoint retired April 30, 2025
Output
32.143 DBU per 1 million output tokens when historically available through Databricks pay-per-token serving; endpoint retired April 30, 2025
Local experimentation, research, instruction-tuning studies, and organizations needing an openly downloadable model with commercial-use licensing
Context
2K
Reasoning
3/10
Speed
4/10
Status
Legacy open-weight model; publicly downloadable; no official shutdown date found
Input
No official hosted API price; model weights are available for self-hosted use
Output
No official hosted API price; model weights are available for self-hosted use
Self-hosted general-purpose text generation, experimentation, and fine-tuning with open model weights.
Context
66K
Reasoning
5/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Open weights; legacy model
Short-form instruction following, local inference, open-model research, and domain fine-tuning
Context
2K
Reasoning
4/10
Speed
7/10
Status
Legacy open-weight model; downloadable and usable for self-hosted deployment
Self-hosted code generation, completion, repository analysis, debugging, code translation, and programming-focused research.
Context
128K
Reasoning
7/10
Speed
5/10
Status
Legacy open-weight model series; downloadable checkpoints remain available, while the hosted Coder API line was superseded and merged into DeepSeek-V2.5.
Local code completion, fill-in-the-middle generation, IDE integrations, repository-level coding, and private or self-hosted inference
Context
131K
Reasoning
6/10
Speed
8/10
Status
Open-weight and downloadable; accessible through the official Hugging Face repository, with no verified provider-published shutdown date
Local code generation, code completion, debugging, code explanation, repository-scale prompts, and developers needing an open-weight coding model
Context
128K
Reasoning
7/10
Speed
7/10
Status
Available as an open-weight downloadable model; older but still accessible
Local bilingual text generation, research, experimentation, instruction tuning, and lightweight self-hosted assistants
Context
4K
Reasoning
5/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Legacy open-weight model; downloadable and usable through compatible local inference tools
Input
No official hosted API price; downloadable weights for self-hosting
Output
No official hosted API price; self-hosting and infrastructure costs apply
Local deployment, bilingual English-Chinese generation, language-model research, mathematics, coding experiments, and custom fine-tuning workflows
Context
4K
Reasoning
6/10
Speed
3/10
Status
Legacy open-weight model; downloadable checkpoints remain available, but it is superseded by newer DeepSeek model families
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; downloadable weights
Advanced mathematical reasoning, natural-language theorem proving, proof generation, proof verification, and research on self-correcting reasoning systems
Context
164K
Reasoning
10/10
Speed
2/10
Status
Available as an open-weight research model; no official DeepSeek-hosted API deployment or public model-specific API pricing verified
Local OCR, document digitization, PDF and image parsing, layout-aware markdown conversion, table extraction, figure parsing, and visual-text compression research
Context
8K
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current open-weight model; publicly available for local and self-hosted inference
Local OCR, scanned-document transcription, layout-aware document parsing, table extraction, and document-to-Markdown workflows
Context
8K
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current open-weight model; self-hosted checkpoint
Lean 4 theorem proving, formal mathematics, automated proof generation, and theorem-proving research
Reasoning
8/10
Speed
5/10
Status
Open-weight research release; legacy predecessor to DeepSeek-Prover-V1.5 and DeepSeek-Prover-V2
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; downloadable weights
Lean 4 proof completion, formal mathematics research, automated theorem proving, and verifier-guided proof search
Context
4K
Reasoning
8/10
Speed
4/10
Status
Open-weight, downloadable, legacy/superseded
Lean 4 theorem proving, formal mathematics, automated proof synthesis, proof-search research, and verifier-guided reasoning
Context
164K
Reasoning
9/10
Speed
2/10
Status
Open-weight and downloadable; currently accessible from the official model repository; no first-party hosted API identified for this exact checkpoint
Lean 4 theorem proving, formal mathematics research, proof completion, proof-search experiments, and open-weight model fine-tuning
Context
4K
Reasoning
8/10
Speed
4/10
Status
Legacy open-weight model; downloadable and usable, but superseded by newer DeepSeek-Prover releases
Mathematical reasoning, coding, technical analysis, research, and self-hosted reasoning applications
Context
128K
Reasoning
9/10
Speed
4/10
Status
Open-weight checkpoint available; original hosted API identity superseded and scheduled for discontinuation
Self-hosted mathematics, coding, research, complex reasoning, and long-form analysis
Context
131K
Reasoning
9/10
Speed
4/10
Status
Available open-weight model
Reasoning research, mathematics, coding experiments, reinforcement-learning studies, open-weight evaluation, and model distillation
Context
131K
Reasoning
9/10
Speed
3/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight and downloadable; experimental/legacy research model; not a current first-party hosted API model
Local mathematical reasoning, compact reasoning experiments, educational applications, lightweight coding assistance, and self-hosted inference on limited hardware
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Active open-weight model; downloadable for self-hosted inference
Local or third-party deployment for general text generation, translation, mathematics, research, and code generation
Context
128K
Reasoning
7/10
Speed
7/10
Status
Legacy open-weight model; downloadable and usable through local or third-party inference, but not listed in DeepSeek's current first-party hosted API catalog
Local text generation, Chinese and English language tasks, MoE research, fine-tuning, and efficient self-hosted inference
Context
33K
Reasoning
5/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight and downloadable; legacy self-hosting model
Input
No official DeepSeek-hosted API price documented for this exact model; self-hosted weights are available under the DeepSeek Model License.
Output
No official DeepSeek-hosted API price documented for this exact model; self-hosted inference costs depend on hardware and serving infrastructure.
Open-weight general language generation, coding assistance, code completion, and self-hosted experimentation
Context
128K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Legacy open-weight model; the V2.5 series was superseded by newer DeepSeek model families, while the model weights remain available
Open-weight general language generation, coding, long-context text tasks, research, and cost-sensitive third-party inference.
Context
128K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Superseded and no longer current as a first-party hosted API model; open-weight checkpoint remains available for self-hosted and third-party deployment.
Input
$0.27 per 1M tokens for cache misses; $0.07 per 1M tokens for cache hits, historical launch pricing
Output
$1.10 per 1M tokens, historical launch pricing
Open-weight reasoning, coding, tool-calling, long-context analysis, and self-hosted agent systems
Context
128K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Legacy hosted API generation; official open-weight release remains available
Open-weight deployment, coding assistance, long-context text processing, reasoning workflows, search agents, and terminal-oriented automation
Context
128K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Open-weight checkpoint available; dedicated DeepSeek API endpoint retired on 2025-10-15
Input
$0.56 per 1M input tokens cache miss; $0.07 per 1M cached input tokens during historical API availability
Output
$1.68 per 1M output tokens during historical API availability
Difficult mathematics, advanced coding, scientific reasoning, long-form analysis, benchmark evaluation, and research deployment
Context
164K
Reasoning
10/10
Speed
4/10
Status
Retired hosted API; open-weight model remains available for self-hosting and third-party deployment
Input
Historical temporary API pricing was the same as DeepSeek-V3.2; no current hosted price because the endpoint expired on 2025-12-15
Output
Historical temporary API pricing was the same as DeepSeek-V3.2; no current hosted price because the endpoint expired on 2025-12-15
Local image understanding, visual question answering, diagram and document analysis, multimodal research, and compact deployments
Context
4K
Reasoning
3/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Available open-weight checkpoint; legacy first-generation model
Open-weight reasoning, coding, long-context analysis, tool-using agents, research workflows, and cost-sensitive deployments
Context
131K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Legacy open-weight model; former DeepSeek API aliases deepseek-chat and deepseek-reasoner were scheduled for discontinuation on 2026-07-24
Input
$0.028 per 1M tokens cached; $0.28 per 1M tokens cache miss during official API availability
Output
$0.42 per 1M tokens during official API availability
Long-context reasoning, coding, agent workflows, and cost-sensitive API applications
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Retired; legacy API identifier temporarily routed to DeepSeek-V4.1-Flash from September 10, 2026
Input
$0.14 per 1 million input tokens
Output
$0.28 per 1 million output tokens
Self-hosted language-model research, custom post-training, domain adaptation, and large-context text generation
Context
1.05M
Reasoning
7/10
Speed
6/10
Status
Current downloadable open-weight base checkpoint; no Hugging Face Inference Provider deployment listed
Image understanding, screenshot and chart analysis, multimodal coding agents, visual tool-use workflows, and text-plus-image reasoning
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Status
Retired as an independent model on 2026-09-10; legacy API identifier temporarily routes requests to DeepSeek-V4.1-Flash
Input
$0.15 per 1M cache-miss input tokens off-peak or $0.30 peak; $0.003 off-peak or $0.006 peak for cache-hit input when using the current routed Flash pricing
Output
$0.60 per 1M output tokens off-peak or $1.20 peak when using the current routed Flash pricing
Complex reasoning, coding agents, long-context analysis, tool-using workflows, and large document or codebase processing
Context
1M
Reasoning
9/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Deprecated for independent serving; deepseek-v4-pro API requests are routed to DeepSeek-V4.1-Flash until V4.1-Pro launches
Input
Official V4-Pro list price: US$0.66 per 1M cache-miss input tokens off-peak and US$1.32 peak; US$0.022 per 1M cache-hit input tokens off-peak and US$0.044 peak. DeepSeek announced that routed requests use V4.1-Flash rates.
Output
Official V4-Pro list price: US$1.98 per 1M output tokens off-peak and US$3.96 peak. DeepSeek announced that routed requests use V4.1-Flash rates.
Research, continued pretraining, fine-tuning, foundation-model evaluation, and custom large-scale inference
Context
1.05M
Reasoning
8/10
Speed
4/10
Status
Current open-weight base checkpoint; downloadable under the MIT License
Low-cost, high-throughput reasoning and coding, long-context analysis, agentic workflows, tool calling, and text-plus-image understanding.
Context
1.05M
Reasoning
9/10
Speed
10/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Current and available through the DeepSeek API; the canonical API identifier is deepseek-flash.
Input
$0.15 per 1M input tokens off-peak or $0.30 peak for cache misses; $0.003 off-peak or $0.006 peak for cache hits.
Output
$0.60 per 1M output tokens off-peak or $1.20 peak.
Local research, image understanding, visual question answering, multimodal prototyping, and lightweight text-to-image experimentation
Context
4K
Reasoning
4/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Available as an open-weight research model; superseded in the Janus series by newer Janus-Pro variants but still downloadable and usable.
Input
No official hosted API pricing; self-hosted model weights
Output
No official hosted API pricing; self-hosted model weights
Local research, visual question answering, image interpretation, and compact text-to-image experimentation
Context
4K
Reasoning
5/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Available as an open-weight downloadable checkpoint; no official hosted inference API identified
Input
No official hosted API pricing
Output
No official hosted API pricing
Browser automation, visual UI interaction, repetitive web workflows, form filling, and user-interface testing
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Legacy preview; currently documented and accessible, with newer Gemini 3.x models recommended for new computer-use applications
Input
$1.25 per 1M tokens for prompts up to 200K tokens; $2.50 per 1M tokens for prompts over 200K tokens
Output
$10.00 per 1M tokens for prompts up to 200K tokens; $15.00 per 1M tokens for prompts over 200K tokens
Large-scale multimodal processing, low-latency reasoning, coding, data extraction, tool-using agents, and applications requiring a very large context window.
Context
1.05M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable and currently served through the Gemini API with restricted access for users who have actively used Gemini 2.5 models; not deprecated; no shutdown date announced.
Input
$0.30 per 1M text/image/video tokens; $1.00 per 1M audio tokens; batch input $0.15 per 1M text/image/video tokens.
Output
$2.50 per 1M tokens, including thinking tokens.
High-volume classification, simple extraction, lightweight multimodal analysis, routing, tagging, summarization, and extremely latency-sensitive applications.
Context
1.05M
Reasoning
6/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable; currently served through the Gemini API with access limited to users who have actively used Gemini 2.5 models. Google recommends newer models for new projects.
Input
$0.10 per 1M text/image/video input tokens; $0.30 per 1M audio input tokens. Batch/Flex: $0.05 per 1M text/image/video tokens and $0.15 per 1M audio tokens.
Output
$0.40 per 1M output tokens. Batch/Flex: $0.20 per 1M output tokens.
Real-time voice and video agents, speech-to-speech assistants, interactive customer support, tutoring, coaching, and multimodal Live API applications
Context
131K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Preview; currently listed by Google with limited access for users who have actively used Gemini 2.5 models; no shutdown date announced
Input
$0.50 per 1M text tokens; $3.00 per 1M audio or video tokens
Output
$2.00 per 1M text tokens; $12.00 per 1M audio tokens
Low-latency controllable text-to-speech, voice assistants, narration, read-aloud features, and multi-speaker audio generation
Context
8K
Reasoning
2/10
Speed
8/10
Status
Preview; currently available with limited access conditions
Input
$0.50 per 1M input text tokens; batch: $0.25 per 1M input text tokens
Output
$10.00 per 1M output audio tokens; batch: $5.00 per 1M output audio tokens
Advanced coding, complex reasoning, mathematics, STEM analysis, long documents, large codebases, multimodal analysis, and tool-using agents.
Context
1.05M
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable and generally available through the Gemini API; access is currently limited to users who have actively used Gemini 2.5 models. Google states that the model is not deprecated and will continue to be served until further notice.
Input
$1.25 per 1M tokens for prompts up to 200K tokens; $2.50 per 1M tokens for prompts over 200K tokens. Batch/Flex rates are $0.625/$1.25 respectively.
Output
$10.00 per 1M tokens for prompts up to 200K tokens; $15.00 per 1M tokens for prompts over 200K tokens. Batch/Flex rates are $5.00/$7.50 respectively. Prices include thinking tokens.
High-fidelity single-speaker and multi-speaker narration, audiobooks, podcasts, professional voiceovers, and scripted creative audio
Context
8K
Reasoning
1/10
Speed
6/10
Status
Preview; currently listed and accessible through the Gemini API, with restricted access to Gemini 2.5 models
Input
$1.00 per 1M text tokens standard; $0.50 per 1M text tokens batch
Output
$20.00 per 1M audio tokens standard; $10.00 per 1M audio tokens batch
Fast image generation, conversational image editing, image transformation, and high-volume visual workflows
Context
66K
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Deprecated; scheduled to shut down on 2026-10-02
Input
$0.30 per 1 million text or image input tokens; Batch/Flex $0.15; Priority $0.54
Output
$0.039 per generated image; Batch/Flex $0.0195; Priority $0.0702
Agentic workflows, everyday coding, reasoning and planning, multimodal analysis, long-context document work, and cost-sensitive tool-using applications
Context
1.05M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
$0.50 per 1 million input tokens
Output
$3 per 1 million output tokens
Complex reasoning, advanced software engineering, long-context multimodal analysis, and agentic workflows requiring reliable tool use
Context
1.05M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
$2.00 per 1M tokens for prompts up to 200K tokens; $4.00 per 1M tokens for prompts over 200K tokens
Output
$12.00 per 1M tokens for prompts up to 200K tokens; $18.00 per 1M tokens for prompts over 200K tokens, including thinking tokens
Fast multimodal applications, coding assistants, long-context document and video analysis, tool-using agents, enterprise workflows, and rapid agentic execution
Context
1.05M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable; previous-generation Flash model; currently accessible; no shutdown date announced
Input
$1.50 per 1 million input tokens
Output
$7.50 per 1 million output tokens
Coding, software engineering, tool-using agents, long-context multimodal analysis, structured extraction and high-throughput enterprise workflows
Context
1.05M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Generally available; previous-generation Flash model, currently supported
Input
$0.75 per 1M tokens through 2026-12-31; $1.50 per 1M tokens starting 2027-01-01
Output
$3.75 per 1M tokens, including thinking tokens, through 2026-12-31; $7.50 per 1M tokens starting 2027-01-01
Long-horizon software engineering, autonomous agents, multimodal document workflows, enterprise knowledge work, and tool-using applications.
Context
1.05M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Generally available
Input
$0.75 per 1 million tokens through December 31, 2026; $1.50 per 1 million tokens from January 1, 2027. Batch/Flex introductory input price: $0.375 per 1 million tokens.
Output
$3.75 per 1 million tokens, including thinking tokens, through December 31, 2026; $7.50 per 1 million tokens from January 1, 2027. Batch/Flex introductory output price: $1.875 per 1 million tokens.
High-volume translation, classification, extraction, summarization, document processing, and lightweight tool-using agent workflows
Context
1.05M
Reasoning
7/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Deprecated; generally available and accessible until scheduled shutdown on May 7, 2027
Input
$0.25 per 1M text/image/video tokens; $0.50 per 1M audio tokens; batch/flex: $0.125 per 1M text/image/video tokens and $0.25 per 1M audio tokens
Output
$1.50 per 1M tokens standard; $0.75 per 1M tokens for batch/flex inference
Low-latency voice agents, real-time dialogue, multimodal live sessions, and interactive audio applications
Context
131K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Legacy preview; currently accessible; Google recommends Gemini 3.8 Live for most new low-latency voice-agent deployments
Input
$0.75 per 1M text tokens; $3.00 per 1M audio tokens or $0.005 per audio minute; $1.00 per 1M image/video tokens or $0.002 per image/video minute
Output
$4.50 per 1M text tokens; $12.00 per 1M audio tokens or $0.018 per audio minute
Controllable expressive speech, narration, accessibility, scripted audio, and multi-speaker TTS prototypes
Context
8K
Reasoning
2/10
Speed
8/10
Status
Legacy preview; currently accessible; no shutdown date announced
Input
$1.00 per 1 million text tokens standard; $0.50 per 1 million text tokens batch
Output
$20.00 per 1 million audio tokens standard; $10.00 per 1 million audio tokens batch
Gemini 3.1 Flash Lite Image
Fast, low-cost 1K image generation and editing, rapid visual prototyping, interactive applications, high-volume image variations, storyboarding, and lightweight creative workflows
Context
66K
Reasoning
3/10
Speed
10/10
Capabilities
Image input
Video input
Multimodal input
Status
Generally available; scheduled retirement June 28, 2027 or later
Input
$0.25 per 1 million input tokens for text, image, and video input; batch pricing $0.125 per 1 million input tokens
Output
$1.50 per 1 million text/thought output tokens and $30 per 1 million image output tokens; approximately $0.0336 per 1K image; batch image output pricing $15 per 1 million image tokens
Agentic workflows, coding agents, long-context multimodal analysis, tool use, and scaled production applications
Context
1.05M
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Generally available; stable
Input
$1.50 per 1M tokens standard; $0.75 per 1M tokens Batch and Flex; $2.70 per 1M tokens Priority
Output
$9.00 per 1M tokens standard; $4.50 per 1M tokens Batch and Flex; $16.20 per 1M tokens Priority
High-volume, latency-sensitive agentic workflows, document parsing, translation, classification, data extraction, search-backed applications, and multimodal sub-agents.
Context
1.05M
Reasoning
8/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
General availability
Input
$0.30 per 1 million tokens for standard text, image, video, and audio input; $0.15 per 1 million input tokens for Batch and Flex; $0.54 per 1 million input tokens for Priority.
Output
$2.50 per 1 million output tokens for standard usage; $1.25 per 1 million output tokens for Batch and Flex; $4.50 per 1 million output tokens for Priority.
Low-latency, real-time speech-to-speech translation for calls, meetings, travel, customer support, and multilingual voice applications
Context
131K
Reasoning
2/10
Speed
10/10
Capabilities
Audio input
Streaming
Input
$3.50 per 1 million input audio tokens, approximately $0.0053 per minute
Output
$21.00 per 1 million output audio tokens, approximately $0.0315 per minute
Pre-recorded audio transcription, multilingual speech recognition, speaker-labeled transcripts, timestamped transcripts, smart dictation, and domain-specific vocabulary
Context
96K
Reasoning
2/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Status
Generally available (GA)
Input
$2.00 per 1M audio input tokens, approximately $0.003 per minute; free tier available
Output
$12.00 per 1M text output tokens, approximately $0.002 per minute; free tier available
High-volume text-to-speech production, low-latency voice-agent cascades, read-aloud applications, voice replication, and everyday single-speaker speech
Context
8K
Reasoning
1/10
Speed
9/10
Status
Generally available
Input
$0.50 per 1 million text tokens through December 31, 2026; $1.00 per 1 million text tokens starting January 1, 2027. Batch and Flex input: $0.25 through December 31, 2026; $0.50 starting January 1, 2027.
Output
$6.00 per 1 million audio tokens through December 31, 2026; $12.00 per 1 million audio tokens starting January 1, 2027. Standard equivalent: approximately $0.0015 per 10 seconds of audio through December 31, 2026.
Studio-quality narration, audiobooks, expressive voice acting, complex multi-speaker dialogue, regional accents, difficult pronunciations, long-form narration, voice design, and voice replication.
Context
8K
Reasoning
1/10
Speed
8/10
Status
Generally available (GA); no shutdown date announced
Input
$0.50 per 1 million text input tokens through December 31, 2026; $1.00 per 1 million text input tokens from January 1, 2027
Output
$9.00 per 1 million audio output tokens through December 31, 2026; $18.00 per 1 million audio output tokens from January 1, 2027. Equivalent to $0.00225 per 10 seconds of audio through December 31, 2026 and $0.0045 per 10 seconds thereafter.
Low-latency voice agents, real-time audio-to-audio dialogue, multimodal assistants, and interactive tool-using applications
Context
131K
Reasoning
7/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable; generally available
Input
$0.75 per 1M text tokens; $3.00 per 1M audio tokens or approximately $0.005 per minute; $1.00 per 1M image/video tokens or approximately $0.002 per minute
Output
$4.50 per 1M text tokens; $12.00 per 1M audio tokens or approximately $0.018 per minute
Complex real-time voice agents, multi-step problem solving, asynchronous tool workflows, technical support, travel coordination, and spoken STEM or coding tutoring
Context
131K
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Stable; generally available
Input
$0.75 per 1M text tokens; $3.00 per 1M audio tokens or $0.005 per audio minute; $1.00 per 1M image/video tokens or $0.002 per minute
Output
$4.50 per 1M text tokens; $12.00 per 1M audio tokens or $0.018 per audio minute
Autonomous market research, due diligence, literature reviews, competitive analysis, source-heavy investigations, and cited research reports
Context
1.05M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
Variable pay-as-you-go pricing based on underlying model usage and tools; no fixed per-input-token price is specified for the agent.
Output
Variable pay-as-you-go pricing based on research depth and generated output; typical total cost is estimated at approximately $1–$3 per standard task.
Comprehensive market research, competitive analysis, due diligence, literature reviews, and source-rich investigative reports
Context
1.05M
Reasoning
10/10
Speed
3/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Preview; currently available through the Interactions API in the Gemini API and Google AI Studio
Input
Pay-as-you-go based on underlying Gemini model inference and tool usage; no fixed model-specific input rate published
Output
Pay-as-you-go based on underlying Gemini model inference and tool usage; no fixed model-specific output rate published
Text semantic search, RAG retrieval, document matching, classification, clustering, and recommendation systems
Context
2K
Reasoning
1/10
Speed
8/10
Status
Current, scheduled for shutdown on 2028-05-14
Input
$0.00015 per 1,000 input tokens for online Vertex AI requests; $0.00012 per 1,000 input tokens for batch requests
Output
No charge for embedding output
Cross-modal semantic search, multimodal RAG, vector retrieval, recommendations, classification, clustering, and indexing mixed text and media collections.
Context
8K
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Generally available
Input
Standard paid tier: text $0.20 per 1M tokens; images $0.45 per 1M tokens or $0.00012 per image; audio $6.50 per 1M tokens or $0.00016 per second; video $12.00 per 1M tokens or $0.00079 per frame. Batch pricing is 50% lower.
Output
No separate output-token price; the model returns embeddings. Output is included in the input-modality pricing structure.
Fast, high-volume image generation and editing, visual iteration, marketing assets, diagrams, infographics, localization, and applications requiring image-search grounding
Context
131K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Web search
Status
Generally available; the stable Gemini API model is gemini-3.1-flash-image
Input
$0.50 per 1 million tokens for text/image input; batch input $0.25 per 1 million tokens
Output
$3 per 1 million text/thinking tokens and $60 per 1 million image-output tokens; approximately $0.067 per 1K image, $0.101 per 2K image, and $0.151 per 4K image; batch image output $30 per 1 million image-output tokens
Professional image generation and editing, complex compositions, product mockups, infographics, branded creative, multilingual localization, and high-fidelity visual prototyping
Context
66K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Multimodal input
Web search
Status
Current stable model
Input
$2.00 per 1M text/image input tokens; approximately $0.0011 per input image
Output
$12.00 per 1M text and thinking tokens; $120.00 per 1M image tokens, equivalent to approximately $0.134 per 1K/2K image and $0.24 per 4K image
Fast text-to-video, image-to-video, conversational video editing, video extension, interpolation, marketing content, and short-form cinematic production
Context
1.05M
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Current; stable model available as gemini-omni-1.1-flash, with gemini-omni-flash-preview also documented
Input
$1.50 per 1M input tokens for text, image, video, and audio inputs under standard pricing
Output
$17.50 per 1M video output tokens; $9.00 per 1M text output tokens where applicable
High-level robot planning, spatial reasoning, video progress tracking, tool orchestration, and multi-robot collaboration
Context
131K
Reasoning
8/10
Speed
4/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
$1.00 per 1 million tokens for the standard preview endpoint through December 31, 2026; $2.00 per 1 million tokens starting January 1, 2027
Output
$5.00 per 1 million tokens, including reasoning tokens, through December 31, 2026; $10.00 per 1 million tokens starting January 1, 2027
Low-latency robotic agents, continuous audio/video monitoring, function-based robot orchestration, warehouse workflows, and multi-robot coordination
Context
131K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
$1.00 per 1M tokens through December 31, 2026; $2.00 per 1M tokens from January 1, 2027
Output
$5.00 per 1M tokens through December 31, 2026; $10.00 per 1M tokens from January 1, 2027
Full-length AI-generated songs, vocal music, instrumental arrangements, songwriting experiments, soundtracks, and image-inspired music creation
Context
131K
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Generally available
Input
$0.08 per full song; no free tier
Output
$0.08 per full song
Interactive instrumental music generation, live musical improvisation, prompt-driven DJ tools, MIDI-controlled experiences, and real-time creative audio applications
Reasoning
1/10
Speed
9/10
Status
Experimental and currently documented through the Gemini API as lyria-realtime-exp
Input
Not publicly listed for this model
Output
Not publicly listed for this model
Generating short music or audio clips
Full-length AI music, soundtrack creation, songwriting, structured compositions, advertising audio, games, and creative production workflows
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Input
$0.08 per full-song generation
Cinematic text-to-video and image-to-video generation, short-form storytelling, storyboarding, advertising concepts, visual effects exploration, and creative previsualization with synchronized audio
Context
1K
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Status
Preview; currently accessible through Google APIs and Google products
Input
$0.40 per second for standard Veo 3.1 video with audio at 720p or 1080p; $0.60 per second at 4K
Output
Generated video with native audio; pricing is charged per generated video second
Fast, high-volume video generation, creative iteration, social content, advertising concepts, and automated production workflows
Capabilities
Image input
Video input
Multimodal input
Status
Generally available on Vertex AI; preview model on the Gemini API
Input
$0.10 per video second at 720p; $0.12 per video second at 1080p; $0.30 per video second at 4K where supported
Output
Video with natively generated audio, priced per generated video second
High-volume text-to-video and image-to-video generation, rapid creative iteration, social content, advertising variations, and cost-sensitive production workflows
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Preview; currently available through the Gemini API and Google Cloud Vertex AI
Input
$0.05 per second for 720p video with audio; $0.08 per second for 1080p video with audio
Output
Video with native audio, priced per generated-video second
Zero-shot forecasting of regularly sampled numerical time series across changing temporal resolutions
Context
4K
Reasoning
1/10
Speed
8/10
Status
Current open-weight model; r1.1 revision available
Input
No hosted API pricing documented; self-hosted open weights
Output
No hosted API pricing documented; self-hosted open weights
Japanese text generation, summarization, classification, extraction, question answering, and Japanese-English translation in legacy IBM enterprise deployments
Context
4K
Reasoning
4/10
Speed
7/10
Status
Deprecated; withdrawn from IBM Software Hub 5.2.0 and removed from watsonx.ai in 2.2.0
Multilingual enterprise question answering, retrieval-augmented generation, summarization, extraction, classification, and text generation in English, German, Spanish, French, and Portuguese
Context
8K
Reasoning
4/10
Speed
3/10
Status
Retired; deprecated January 15, 2025 and withdrawn from standard watsonx.ai availability April 16, 2025
Fine-tuning, domain adaptation, text generation, summarization, extraction, classification, and question answering
Context
4K
Reasoning
4/10
Speed
7/10
Status
Available for deployment on demand in IBM watsonx.ai; also available as an Apache 2.0 open-weight model through IBM's Hugging Face organization.
Input
$0.0006 per 1,000 tokens
Output
$0.0006 per 1,000 tokens
Fine-tuning, long-context document processing, enterprise text classification, extraction, summarization, question answering, and self-managed deployment
Context
131K
Reasoning
5/10
Speed
6/10
Status
Available; base model intended for fine-tuning and dedicated deployment
Self-hosted enterprise assistants, long-context document analysis, RAG, summarization, extraction, multilingual text workflows, and function calling
Context
131K
Reasoning
5/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available as an open-weight model; superseded by Granite-3.3-8B-Instruct for newer deployments
Long-context enterprise RAG, document and meeting summarization, information extraction, multilingual dialogue, code-related tasks, function calling, and self-hosted deployments
Context
131K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available; legacy in some IBM watsonx catalogs
Local or dedicated enterprise assistants, RAG, long-document summarization, multilingual text workflows, coding assistance, reasoning, and resource-conscious inference
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available; open-weight model with local deployment and dedicated IBM watsonx deployment options
Self-hosted enterprise assistants, long-context RAG, summarization, multilingual text tasks, coding assistance, function calling, and cost-sensitive deployments.
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Retired from IBM watsonx.ai deploy-on-demand service on 2026-02-22; open-weight checkpoint remains available for self-hosted and compatible third-party deployments.
Input
$0.0002 per 1,000 tokens, historical IBM watsonx Developer Hub listing; no current IBM-hosted price verified after withdrawal.
Output
$0.0002 per 1,000 tokens, historical IBM watsonx Developer Hub listing; no current IBM-hosted price verified after withdrawal.
Low-latency local inference, long-context text generation, RAG, multilingual assistants, coding assistance, and lightweight tool-calling agents
Context
128K
Reasoning
5/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Status
Current open-weight instruct model
Enterprise RAG, multi-tool agents, function calling, customer-support automation, multilingual instruction following, and long-context workloads
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Status
Available; open-weight instruct model
Input
$0.0000636 per 1,000 tokens on IBM watsonx.ai multitenant hardware
Output
$0.000265 per 1,000 tokens on IBM watsonx.ai multitenant hardware
Efficient enterprise assistants, multilingual text generation, RAG, classification, extraction, summarization, coding assistance, fill-in-the-middle completion, structured JSON, and tool-calling workflows
Context
128K
Reasoning
5/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current; open-weight instruct model; available for download and listed for deploy-on-demand use in IBM watsonx.ai
Efficient local or private deployment, multilingual enterprise text processing, RAG, summarization, extraction, coding assistance, function calling, and lightweight AI assistants.
Context
131K
Reasoning
5/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Currently available open-weight model; earlier-generation Granite model superseded by the newer Granite 4.2 family, with no verified deprecation or shutdown date.
Input
No official hosted API token price published; downloadable weights are available under Apache 2.0.
Output
No official hosted API token price published; downloadable weights are available under Apache 2.0.
Self-hosted enterprise assistants, multilingual text generation, RAG, coding assistance, structured extraction, and tool-calling agents
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight instruction model
Input
Not available for official per-token hosted pricing
Output
Not available for official per-token hosted pricing
Self-hosted enterprise assistants, long-context RAG, multilingual applications, coding, structured extraction, and tool-calling agents
Context
131K
Reasoning
7/10
Speed
5/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight instruct model; publicly available
Efficient reasoning, coding assistance, tool calling, multilingual dialogue, local deployment, and lightweight enterprise agents
Context
131K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Status
Current; publicly available open-weight model
Local or self-hosted reasoning, coding assistants, tool calling, multilingual dialogue, retrieval-augmented generation, and agentic workflows
Context
131K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current; open-weight and downloadable
Input
No official hosted API price; downloadable Apache 2.0 weights
Output
No official hosted API price; inference infrastructure costs depend on deployment
Self-hosted enterprise reasoning, coding agents, multilingual applications, tool calling, long-context workflows, and organizations requiring Apache 2.0 licensing
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Status
Current; open-weight and available for download
Input
No official IBM hosted API price; model weights are available for self-hosted deployment
Output
No official IBM hosted API price; model weights are available for self-hosted deployment
Self-hosted English text generation, summarization, extraction, classification, experimentation with LAB-aligned open-weight models, and resource-conscious deployments
Context
8K
Reasoning
5/10
Speed
7/10
Status
Withdrawn from IBM watsonx.ai on 2025-01-07; open-weight checkpoint remains available for self-hosted deployment
English enterprise chat, retrieval-augmented generation, question answering, summarization, extraction, and classification
Context
8K
Reasoning
4/10
Speed
6/10
Status
Withdrawn; access ended January 19, 2025
Input
$0.0006 per 1,000 input tokens, historical watsonx.ai rate
Output
$0.0006 per 1,000 output tokens, historical watsonx.ai rate
Self-hosted coding assistants, code generation, code explanation, code conversion, repository-scale prompts, and historical or reproducible research
Context
128K
Reasoning
4/10
Speed
8/10
Status
Retired; withdrawn from IBM watsonx.ai on 2025-07-17
Input
$0.0006 per 1,000 input tokens historically on IBM watsonx.ai
Output
$0.0006 per 1,000 output tokens historically on IBM watsonx.ai
Local code generation, explanation, repair, translation, and research with an open Apache 2.0 model
Context
4K
Reasoning
5/10
Speed
7/10
Status
Deprecated; still publicly available for historical and scientific use
Schema linking and relevant-table or relevant-column selection in text-to-SQL pipelines
Context
8K
Reasoning
3/10
Speed
3/10
Status
Available; deploy on demand only
Input
Not available as pay-as-you-go token pricing; deploy-on-demand infrastructure pricing applies
Output
Not available as pay-as-you-go token pricing; deploy-on-demand infrastructure pricing applies
Natural-language-to-SQL generation over structured databases, analytics assistants, and the SQL-generation stage of text-to-SQL pipelines.
Context
8K
Reasoning
5/10
Speed
4/10
Status
Available; deploy-on-demand model in IBM watsonx.ai
Input
Not publicly listed as a standalone token price; IBM lists the model as deploy-on-demand.
Output
Not publicly listed as a standalone token price; IBM lists the model as deploy-on-demand.
Self-hosted coding assistants, code generation, code conversion, code explanation, and programming experiments where an Apache 2.0 open-weight model is preferred.
Context
8K
Reasoning
5/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight model and documented for IBM watsonx.ai deploy-on-demand use; deprecated and withdrawn from the watsonx.ai multitenant offering according to IBM's 2025 lifecycle notice.
Self-hosted code generation, code explanation, code repair, code conversion, coding assistants, and legacy Granite Code compatibility
Context
8K
Reasoning
6/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Legacy/deprecated; available in some IBM watsonx environments and retained for historical or scientific use
Fast, low-footprint English semantic search, RAG retrieval, similarity matching, and vector indexing
Context
512
Reasoning
1/10
Speed
9/10
Status
Legacy; downloadable and usable, but superseded by Granite Embedding Small English R2
Low-latency multilingual semantic search, retrieval-augmented generation, vector search, document similarity, long-document retrieval, and cross-lingual code retrieval
Context
33K
Reasoning
2/10
Speed
9/10
Status
Current; open-weight model
Low-cost multilingual semantic search, cross-lingual retrieval, vector databases, similarity matching, and RAG pipelines
Context
512
Reasoning
1/10
Speed
9/10
Status
Retired from IBM watsonx.ai; open-weight checkpoint remains available through IBM's Hugging Face organization
Input
No official hosted API price verified; open-weight model intended for local or third-party deployment
Output
Not applicable to token generation; produces 384-dimensional embeddings
English semantic search, vector retrieval, RAG, similarity matching, enterprise knowledge-base search, and local embedding deployment
Context
512
Reasoning
1/10
Speed
8/10
Status
Legacy/superseded but downloadable and usable
Multilingual semantic search, retrieval-augmented generation, cross-lingual retrieval, long-document search, similarity, and code retrieval
Context
33K
Reasoning
1/10
Speed
7/10
Status
Current open-weight model
English semantic search, vector retrieval, retrieval-augmented generation, document similarity, clustering, and enterprise information retrieval
Context
8K
Reasoning
1/10
Speed
8/10
Status
Current open-weight model
Input
No official hosted API price; downloadable weights are available under the Apache 2.0 license
Output
No official hosted API price; downloadable weights are available under the Apache 2.0 license
Compact English semantic search, retrieval-augmented generation, document similarity, and private vector-search deployments
Status
Current open-weight model
Multilingual semantic search, vector retrieval, RAG, clustering, similarity matching, and text classification features
Context
512
Reasoning
1/10
Speed
8/10
Status
Retired; shutdown completed 2026-08-08
Above-ground biomass mapping, forest monitoring, carbon-stock estimation, ecological analysis, and satellite-based remote-sensing research
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight model
Canopy-height mapping, forest monitoring, vegetation analysis, carbon-cycle research, ecological assessment, and remote-sensing experimentation.
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Fine-tuning
Status
Available as downloadable open-weight model; not deployed by an inference provider; IBM repository disclosure states that the project is not maintained as an IBM product.
Land surface temperature estimation, urban heat island analysis, satellite-based environmental monitoring, and temporal gap filling
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available as an open-weight model for local inference
Remote-sensing feature extraction, Earth-observation transfer learning, and regional flood-segmentation workflows using multispectral and SAR satellite imagery
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available as an open-weight research model; no hosted inference provider is currently listed
Input
No official hosted API pricing; open-weight model
Output
No official hosted API pricing; open-weight model
Spatial downscaling of weather forecasts, reanalysis data, and climate simulations
Reasoning
0/10
Speed
0/10
Status
Current and available as downloadable open-weight model
Prompt and response safety classification, jailbreak detection, RAG groundedness and relevance checks, hallucination evaluation, and enterprise AI guardrails
Context
131K
Reasoning
5/10
Speed
7/10
Status
Deprecated in IBM watsonx.ai documentation; open-weight model and official model materials remain available
Input
$0.0002 per 1,000 input tokens in IBM watsonx.ai documentation
Output
$0.0002 per 1,000 output tokens in IBM watsonx.ai documentation
AI safety guardrails, jailbreak detection, RAG groundedness and relevance checks, function-call hallucination detection, custom criteria evaluation, and best-of-N response ranking
Context
8K
Reasoning
6/10
Speed
7/10
Status
Current and available as an open-weight model
Input
Not applicable; no official hosted token price found for the downloadable model
Output
Not applicable; no official hosted token price found for the downloadable model
Multilingual speech-to-text with speaker labels, word-level timestamps, keyword biasing, and self-hosted enterprise transcription
Context
4K
Reasoning
2/10
Speed
7/10
Capabilities
Audio input
Multimodal input
Status
Current open-weight model
Multilingual offline transcription, speech-to-text, speech translation, subtitle generation, and domain-specific recognition with keyword biasing
Context
128K
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Status
Current open-weight model
High-throughput multilingual speech transcription where low inference latency is more important than maximum recognition accuracy.
Context
4K
Reasoning
2/10
Speed
10/10
Status
Current; open-weight; Apache 2.0
Input
No official hosted API pricing; model weights are available for self-hosted use.
Output
No official hosted API pricing; model weights are available for self-hosted use.
High-throughput, low-latency English speech-to-text transcription on local, edge, and enterprise systems
Reasoning
1/10
Speed
10/10
Capabilities
Audio input
Fine-tuning
Streaming
Status
Current; open-weight model
Fast local English speech-to-text transcription, edge-device ASR, research, browser demos, and high-throughput noncommercial batch processing
Reasoning
1/10
Speed
10/10
Capabilities
Audio input
Streaming
Status
Current; research and noncommercial use only
Input
No official hosted API price; downloadable model weights
Output
No official hosted API price; downloadable model weights
Lightweight multivariate forecasting at minute- and hour-level resolutions, especially when 512 historical observations and a 96-point forecast horizon are appropriate.
Context
512
Reasoning
0/10
Speed
9/10
Input
$0.13 per 1,000 input data points
Output
$0.38 per 1,000 output data points
Granite TimeSeries PatchTST-FM
Zero-shot forecasting of demand, prices, energy loads, traffic, telemetry, and other regularly sampled numerical time series; probabilistic forecasts and uncertainty intervals; local or self-managed inference.
Context
8K
Reasoning
0/10
Speed
8/10
Status
Current; open-weight model
High-throughput multivariate time-series forecasting, zero-shot or few-shot forecasting, probabilistic prediction, exogenous-variable forecasting, and CPU-friendly production deployment.
Reasoning
0/10
Speed
9/10
Status
Current open-weight model family; available through the IBM Granite Hugging Face repository. No hosted inference provider deployment was listed on the reviewed model page.
Efficient multivariate forecasting for regularly sampled energy, traffic, manufacturing, network, sales, and sensor data
Context
1K
Reasoning
1/10
Speed
9/10
Status
Available through IBM watsonx.ai; downloadable model branch
Input
watsonx.ai API pricing class 14
Output
watsonx.ai API pricing class 15
Multivariate forecasting when at least 1,536 historical observations per channel are available, especially demand, traffic, electricity, manufacturing, finance, and other minute- or hour-level forecasting tasks.
Context
2K
Reasoning
0/10
Speed
9/10
Input
$0.00013 per 1,000 data points
Output
$0.00038 per 1,000 data points
Chart extraction, table parsing, semantic key-value extraction, visual document processing, and local enterprise RAG pipelines
Reasoning
5/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Current and downloadable; newer Granite Vision 4.1 4B version available
Input
No official hosted API price found; open-weight model intended for self-hosted deployment
Output
No official hosted API price found; open-weight model intended for self-hosted deployment
Enterprise document understanding, chart and table extraction, OCR-oriented image analysis, visual question answering, and multimodal RAG
Context
131K
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available; legacy relative to Granite Vision 4.0 3B Vision
Structured extraction from charts, tables, invoices, forms, and enterprise document images
Reasoning
3/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Streaming
Status
Current open-weight model
Input
No official hosted API price specified; downloadable open-weight model
Output
No official hosted API price specified; downloadable open-weight model
English semantic search, dense retrieval, vector indexing, duplicate-question matching, and lightweight retrieval-augmented generation
Status
Deprecated; withdrawn in Dallas on 2026-09-08 and scheduled for withdrawal in other listed regions on 2027-01-12.
Input
USD 0.0001 per 1,000 input tokens on watsonx.ai
Output
Not separately priced; the model returns embedding vectors rather than generated output tokens
English semantic search, vector database indexing, retrieval-augmented generation, document matching, and query-passage retrieval
Context
512
Reasoning
1/10
Speed
7/10
Status
Deprecated; scheduled for withdrawal on 2027-01-12
Input
$0.0001 per 1,000 input tokens
Bilingual English-Korean chat, instruction following, Korean-language applications, local inference, and open-weight LLM research
Context
4K
Reasoning
7/10
Speed
7/10
Status
Available open-weight model; older EXAONE generation and superseded by newer EXAONE releases
Input
No official hosted API pricing published; self-hosted weights
Output
No official hosted API pricing published; self-hosted weights
Lightweight local text generation, English-Korean assistants, summarization, rewriting, classification, and deployment on resource-constrained devices
Context
33K
Reasoning
5/10
Speed
8/10
Status
Current downloadable open-weight model; research use permitted, commercial use requires a separate license
Input
No official hosted API pricing; downloadable weights
Output
No official hosted API pricing; downloadable weights
Bilingual English-Korean assistants, long-document processing, local deployment, coding support, summarization, translation, and research
Context
33K
Reasoning
7/10
Speed
7/10
Status
Available as an open-weight research model
Local or self-hosted English-Korean text generation, instruction following, long-context processing, coding, mathematics, and organizations requiring downloadable model weights.
Context
33K
Reasoning
7/10
Speed
4/10
Status
Available open-weight model; older EXAONE generation
On-device assistants, local text generation, Korean and multilingual conversational applications, lightweight reasoning, and resource-constrained deployments
Context
66K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current open-weight model
Self-hosted Korean, English, and Spanish language tasks requiring a combination of general-purpose generation, reasoning, coding, long-context processing, or agentic tool use
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Available open-weight model; no official deprecation or shutdown date found
Input
No official hosted API pricing found; intended primarily for self-hosted deployment
Output
No official hosted API pricing found; intended primarily for self-hosted deployment
Open-weight multimodal reasoning, Korean-language tasks, document understanding, OCR, visual question answering, and self-hosted deployments
Context
262K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Current open-weight model
Local mathematics, coding, reasoning experiments, compact bilingual English-Korean applications, and resource-conscious deployment
Context
33K
Reasoning
7/10
Speed
8/10
Status
Current and publicly downloadable open-weight model
Local research, mathematical reasoning, science problem solving, coding evaluation, Korean and English text generation, and self-hosted experimentation
Context
33K
Reasoning
8/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Current and accessible open-weight research release
Mathematical reasoning, scientific problem solving, coding evaluation, and local research deployments
Context
33K
Reasoning
8/10
Speed
5/10
Status
Current open-weight research model; downloadable and locally deployable
Input
No official hosted API price; downloadable weights for local or third-party deployment
Output
No official hosted API price; downloadable weights for local or third-party deployment
Zero-shot probabilistic forecasting of financial and other real-valued time series, especially research and educational workflows involving uncertainty estimates.
Context
512
Reasoning
1/10
Speed
8/10
Status
Current open-weight research release
Whole-slide digital pathology research, molecular subtyping, and image-based mutation prediction
Reasoning
1/10
Speed
4/10
Capabilities
Image input
Fine-tuning
Status
Available research release; a later EXAONE Path 2.0 model has also been released
Computational pathology research, whole-slide image representation learning, cancer biomarker prediction, and biomedical image analysis
Capabilities
Image input
Fine-tuning
Status
Current open-source research model
Research prediction of EGFR mutation status from lung adenocarcinoma whole-slide pathology images
Reasoning
0/10
Speed
7/10
Status
Current research release; gated Hugging Face access; locally deployable
Research on EGFR mutation prediction, computational pathology, molecular subtyping, and whole-slide image biomarker analysis
Reasoning
0/10
Speed
0/10
Status
Available as a gated open-source research release; not deployed by a hosted inference provider
Input
No official hosted API pricing published
Output
No official hosted API pricing published
In-context classification and regression on structured tabular datasets, especially low-retraining workflows and field data with missing values
Reasoning
5/10
Speed
8/10
Status
Current; released for research and education
Long-context reasoning, multilingual generation, Korean-language workloads, coding, tool-using agents, research and distributed self-hosted inference
Context
262K
Reasoning
9/10
Speed
7/10
Status
Current open-weight model
Self-hosted multilingual reasoning, Korean-language applications, coding, long-context document analysis, and tool-enabled agents
Context
256K
Reasoning
8/10
Speed
5/10
Status
Current open-weight model
Image embeddings, dense feature extraction, image retrieval, classification, segmentation, depth estimation, object discovery, video tracking pipelines, and geospatial computer vision
Reasoning
0/10
Speed
7/10
Capabilities
Image input
Fine-tuning
Status
Current; downloadable open-weight research model suite
Local inference, fine-tuning, private deployment, text generation, retrieval-augmented generation, research, and cost-sensitive applications
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight static model; downloadable and usable through compatible local or hosted inference deployments
High-quality open-weight research, multilingual applications, coding, reasoning, synthetic-data generation, model distillation and self-hosted deployments with substantial infrastructure
Context
131K
Reasoning
9/10
Speed
3/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available as an open-weight model; older generation superseded by newer Llama releases
Private local inference, mobile and edge assistants, summarization, rewriting, retrieval-supported generation, and lightweight multilingual applications
Context
128K
Reasoning
3/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available downloadable open-weight model; static checkpoint
Input
No official Meta-hosted API price; downloadable weights are available under the Llama 3.2 Community License
Output
No official Meta-hosted API price; deployment and inference costs depend on hardware or third-party hosting
Private local inference, multilingual text generation, edge applications, model adaptation, and fine-tuning
Context
128K
Reasoning
5/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Status
Available; static pretrained open-weight model
Input
No official Meta-hosted API price; downloadable weights
Output
No official Meta-hosted API price; downloadable weights
Visual question answering, image reasoning, chart and document understanding, image captioning, multimodal research, and self-hosted or partner-hosted AI applications.
Context
128K
Reasoning
8/10
Speed
4/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Available open-weight model; static model trained on an offline dataset
Self-hosted visual question answering, image captioning, document analysis, visual reasoning, and multimodal assistants
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Available open-weight static model
Self-hosted or hosted multilingual chat, coding assistance, long-context text generation, tool calling, synthetic data, and applications requiring open model weights
Context
128K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Status
Available open-weight model; static model trained on an offline dataset
Input
No universal Meta-hosted per-token price; downloadable weights and third-party hosting options are available
Output
No universal Meta-hosted per-token price; downloadable weights and third-party hosting options are available
Open-weight multimodal assistants, image understanding, visual question answering, coding, multilingual applications, creative writing and long-context text processing.
Context
1M
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Available; open-weight static checkpoint
Long-context document and code analysis, visual question answering, multimodal assistants, multilingual applications, self-hosted inference, and customized deployments
Context
10M
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Available; open-weight model
Text and image moderation for prompts and generated responses in generative AI systems
Context
8K
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current and available as an open-weight model
Low-cost prompt and response safety classification, local moderation, mobile and edge deployments, and customizable LLM guardrails
Context
131K
Reasoning
3/10
Speed
9/10
Status
Current open-weight model; downloadable subject to access approval
Self-hosted multilingual input and output moderation for LLM applications
Context
131K
Reasoning
3/10
Speed
6/10
Status
Available open-weight safety classifier
Safety classification of mixed text-and-image prompts and text responses in multimodal LLM systems
Context
128K
Reasoning
4/10
Speed
5/10
Capabilities
Image input
Multimodal input
Status
Available open-weight multimodal safety model; Meta's current model repositories continue to list it.
Low-latency detection of prompt injections and jailbreak attempts in LLM applications, agents, retrieved documents, and other untrusted text
Context
512
Reasoning
1/10
Speed
9/10
Status
Current open-weight safety classifier
Multilingual prompt-injection detection, jailbreak screening, agent security, and filtering untrusted text before it reaches an LLM
Context
512
Reasoning
1/10
Speed
8/10
Status
Current open-weight model; gated download access
Local agents, long-running tool workflows, coding assistants, multimodal document and screenshot understanding, private on-device inference, and model customization
Context
131K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; self-hosted and available through selected third-party hosted inference providers
Input
No official Meta-hosted API token price; open weights can be downloaded and self-hosted without per-token charges
Output
No official Meta-hosted API token price; third-party hosted inference pricing varies by provider
Text-to-image generation, precise image editing, multi-image composition, anchored visual series, product imagery, creative assets, and grounded visual content.
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Status
Current and available through Meta Model API; also available in selected Meta AI consumer experiences
Output
$0.01 per successfully generated image
Agentic workflows, coding agents, computer-use automation, multimodal document and media analysis, long-context reasoning, tool orchestration, and web-grounded applications.
Context
1.05M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current but superseded by Muse Spark 1.2 and Muse Spark 1.3; available on the Meta Model API Standard tier in public preview for US developers.
Input
$1.25 per 1 million input tokens; cached input $0.15 per 1 million tokens
Output
$4.25 per 1 million output tokens
Long-horizon coding agents, repository-scale software engineering, multimodal code generation, debugging, refactoring, and tool-driven workflows
Context
1.05M
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Available; previous version, with Muse Spark 1.3 recommended for new work
Input
$1.25 per 1 million input tokens; $0.15 per 1 million cached input tokens
Output
$4.25 per 1 million output tokens
Long-horizon coding agents, software engineering, browser and computer-use workflows, tool orchestration, large repositories, document analysis, and multimodal reasoning
Context
1.05M
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; available through Meta Model API and Muse Code
Input
$1.25 per 1M input tokens; $0.15 per 1M cached input tokens
Output
$4.25 per 1M output tokens
Real-time speech-to-text, live captions, voice agents, meeting transcription, call intelligence, dictation, and speaker-aware transcription
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Streaming
Status
Current and available through Meta Model API
Input
$3.00 per 1,000 minutes ($0.18 per hour)
Multilingual speech representation learning, audio embeddings, low-resource language research, and custom downstream speech systems
Reasoning
1/10
Speed
5/10
Capabilities
Audio input
Fine-tuning
Status
Current; open-source research model family
Input
Free to download and self-host; no official hosted API price published
Output
Free to download and self-host; no official hosted API price published
Cross-modal audio-video-text retrieval, audiovisual embeddings, sound-event understanding, media indexing, and multimodal perception systems.
Reasoning
2/10
Speed
7/10
Capabilities
Audio input
Video input
Multimodal input
Status
Current and openly available
Input
No official hosted API pricing; open checkpoints are available for self-managed use.
Output
No official hosted API pricing; the model returns embeddings rather than generated media.
Single-image reconstruction of textured 3D objects from natural scenes, 3D computer-vision research, Gaussian-splat workflows, and rapid asset prototyping
Reasoning
2/10
Speed
3/10
Status
Available research release; gated model checkpoints
Prompted audio separation, speech and noise isolation, instrument and vocal extraction, audiovisual sound segmentation, and audio-editing research
Reasoning
2/10
Speed
7/10
Capabilities
Audio input
Video input
Multimodal input
Status
Current open research release; downloadable checkpoints and public demo available
Input
No hosted API pricing published; downloadable checkpoints available under the SAM License
Output
No hosted API pricing published; model outputs separated target and residual audio
Reference-free evaluation and benchmarking of text-guided audio-separation outputs
Reasoning
2/10
Speed
4/10
Capabilities
Audio input
Multimodal input
Status
Current gated research release
Multilingual automatic speech recognition, speech-to-text translation, text translation, text-to-speech translation, and speech-to-speech translation
Reasoning
1/10
Speed
5/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Status
Available open-weight research model; noncommercial research use
Input
Not applicable; no official hosted API pricing published
Output
Not applicable; no official hosted API pricing published
Open-vocabulary object detection, pixel-level image segmentation, and multi-object video tracking
Reasoning
2/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current and available; hosted through Meta Model API and available as released research checkpoints
Input
$2.50 per 1,000 images; $0.20 per 1,000 video frames
Output
Included in the image and video segmentation pricing; no separate output-token price
Computational neuroscience, fMRI response prediction, brain encoding, multisensory research, and in-silico experiment design
Reasoning
0/10
Speed
0/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Fine-tuning
Status
Open-weight research release
Text-guided biomedical image segmentation, annotation assistance, organ and tumor delineation, pathology-cell analysis, and research-oriented medical imaging pipelines
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current; open-weight research model and available for managed deployment through Microsoft Foundry
Input
Not publicly listed as a per-inference model price; Microsoft Foundry deployments incur managed Azure compute charges
Output
Not publicly listed as a per-inference model price; Microsoft Foundry deployments incur managed Azure compute charges
First-pass chest X-ray report drafting, structured findings extraction, radiologist workflow assistance, research evaluation, and institution-specific fine-tuning
Reasoning
2/10
Speed
9/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Limited preview; registration and eligibility approval required
Input
$0.00218 per image for standard inference
Output
Included in the per-image inference price; output is generated text findings
Local image captioning, object detection, phrase grounding, region description, OCR and lightweight computer-vision pipelines
Context
1K
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight model; downloadable from Hugging Face and usable for local or self-hosted inference
Local image captioning, object detection, visual grounding, OCR, region annotation and multi-task computer-vision pipelines.
Context
1K
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight model; publicly accessible through Hugging Face for local or self-hosted inference.
Open-weight reasoning, mathematics, coding, research, and general text-generation applications requiring DeepSeek-R1-style reasoning with Microsoft post-training
Context
164K
Reasoning
8/10
Speed
3/10
Status
Available as an open-weights model and through Microsoft Foundry hosted API
High-quality text-to-image generation, photorealistic imagery, product and marketing visuals, presentation graphics, and precise image-to-image editing.
Context
131K
Reasoning
6/10
Speed
6/10
Capabilities
Image input
Multimodal input
Input
$5 per 1 million text input tokens; $8 per 1 million image input tokens
Output
$47 per 1 million image output tokens
High-quality text-to-image generation, controlled image editing, commercial imagery, product and branding visuals, photorealistic scenes, and multi-reference creative workflows
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Web search
Fast, cost-conscious text-to-image generation, image editing, creative production, concept visualization, and high-volume image workflows
Context
32K
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Multimodal input
Status
Public preview; scheduled for retirement on 2026-10-01
Input
$1.75 per 1M text input tokens; $1.75 per 1M image input tokens
Output
$19.50 per 1M image output tokens
High-fidelity text-to-image generation, precise image editing, hero imagery, commercial and photorealistic creative work, accurate in-image typography, and visually dense scenes requiring consistent objects, characters, materials, and spatial relationship
Context
131K
Reasoning
7/10
Speed
4/10
Capabilities
Image input
Multimodal input
Input
$5 per 1 million text input tokens; $8 per 1 million image input tokens
Output
$106 per 1 million image output tokens
Fast, high-volume text-to-image generation, image editing, marketing assets, product imagery and production design
Context
32K
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Multimodal input
Web search
Complex mathematical reasoning, software engineering, quantitative enterprise analysis, long-context document work, and application-managed agent workflows.
Context
256K
Reasoning
9/10
Speed
7/10
Capabilities
Tool use
Streaming
Multilingual speech-to-text, captions, meeting transcription, accessibility, call analysis, content workflows, voice-agent audio understanding, and domain-specific terminology
Reasoning
1/10
Speed
9/10
Status
Preview; currently accessible through Microsoft Foundry and Azure Speech
Input
$0.36 per hour of audio
Multilingual audio transcription, meeting and contact-center records, captions, clinical notes, accessibility, media search, voice-agent evaluation, and domain-specific transcription with speaker labels and timestamps
Input
$0.10 per hour of audio; limited-time launch pricing through December 31, 2026
Output
Included; no separate text-output charge documented
Expressive long-form narration, audiobooks, podcasts, educational content, voice-over, accessibility, and high-fidelity branded audio
Reasoning
1/10
Speed
7/10
Capabilities
Audio input
Multimodal input
Input
$22 per 1 million characters
Output
$22 per 1 million characters
Low-latency expressive speech for voice agents, assistants, call centers, IVR systems, and interactive multilingual applications
Input
$15 per 1 million characters
Output
$15 per 1 million characters
Medical image and text embeddings, similarity search, multimodal retrieval, downstream classification, outlier detection, dataset curation, and healthcare AI development workflows
Capabilities
Image input
Multimodal input
Fine-tuning
Text-prompted segmentation of complete CT or MRI volumes, organ and lesion delineation, volumetry, annotation assistance, and medical-imaging research
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Multimodal input
Status
Available through Microsoft Foundry classic managed compute; research and model-development use only
Microsoft Foundry model-router
Enterprise applications with mixed-complexity workloads, model selection automation, agent workflows, cost optimization, and configurable quality-versus-latency trade-offs.
Context
200K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Current; latest documented version is 2025-11-18
Input
Not publicly listed as a numeric rate; Microsoft’s pricing page currently shows $- and directs users to Azure pricing for applicable deployment details.
Output
Not separately listed; billing depends on model-router and the underlying routed model pricing configuration.
Local and private text generation, coding assistance, mathematics, reasoning, summarization, and latency-sensitive applications
Context
4K
Reasoning
7/10
Speed
7/10
Status
Retired from Microsoft Foundry; open-weight repository remains available for independent deployment
Input
$0.00017 per 1,000 tokens historically on Azure AI; hosted price no longer applicable after retirement
Output
$0.00068 per 1,000 tokens historically on Azure AI; hosted price no longer applicable after retirement
Long-context chat, retrieval-augmented generation, document summarization, coding, mathematics, reasoning, and self-hosted text-generation applications
Context
131K
Reasoning
7/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Generally available in Microsoft Foundry; open-weight checkpoint available for download and self-hosting
Input
$0.17 per 1 million input tokens in Microsoft Foundry
Output
$0.68 per 1 million output tokens in Microsoft Foundry
Local or low-latency text generation, lightweight assistants, mathematics, coding, summarization, and resource-constrained deployments
Context
4K
Reasoning
6/10
Speed
9/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Microsoft Foundry on August 30, 2025
Input
$0.00013 per 1,000 input tokens (historical Azure pricing)
Output
$0.00052 per 1,000 output tokens (historical Azure pricing)
Long-document analysis, local assistants, code and math tasks, retrieval-augmented generation, private or offline inference, and resource-constrained deployments.
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Microsoft Foundry on August 30, 2025; open-weight checkpoint remains downloadable and usable for self-hosted inference.
Input
$0.00013 per 1,000 input tokens, historical Azure Models-as-a-Service pricing; hosted offering retired
Output
$0.00052 per 1,000 output tokens, historical Azure Models-as-a-Service pricing; hosted offering retired
Local and private text generation, instruction following, code and mathematics assistance, summarization, extraction, and resource-constrained deployments
Context
8K
Reasoning
7/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Microsoft Foundry on 2025-08-30; downloadable open-weight checkpoint remains available
Input
$0.00015 per 1,000 input tokens (historical Azure Models as a Service pricing; hosted service retired)
Output
$0.0006 per 1,000 output tokens (historical Azure Models as a Service pricing; hosted service retired)
Long-context text generation, document analysis, summarization, local assistants, coding support, mathematics, and cost-sensitive self-hosted applications
Context
131K
Reasoning
6/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Microsoft Foundry on 2025-08-30; open-weight checkpoint remains available for self-hosted deployment
Input
USD 0.00015 per 1,000 input tokens historically on Azure Foundry; hosted Azure access is retired
Output
USD 0.0006 per 1,000 output tokens historically on Azure Foundry; hosted Azure access is retired
Local multilingual chat, summarization, document analysis, coding assistance, long-context retrieval, and resource-constrained deployments
Context
131K
Reasoning
6/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model; also supported in Microsoft and third-party local or hosted inference environments
Long-context multilingual assistants, coding, mathematics, reasoning, retrieval-augmented generation, and controlled local deployment.
Context
131K
Reasoning
7/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Retired from Azure Foundry on August 30, 2025; downloadable open-weight checkpoint remains available for self-hosted deployment.
Input
$0.16 per 1M input tokens historically on Azure; current hosted pricing unavailable after retirement.
Output
$0.64 per 1M output tokens historically on Azure; current hosted pricing unavailable after retirement.
Lightweight image and text reasoning, OCR, charts, tables, diagrams, documents, screenshots, and multi-image comparison
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Streaming
Status
Current and accessible; open-weight model and available through Microsoft Foundry deployment options
Efficient local or hosted text generation, STEM reasoning, mathematics, coding assistance, technical question answering, and research on small language models
Context
16K
Reasoning
8/10
Speed
8/10
Status
Preview in Microsoft Foundry; open-weight model available under the MIT license
Efficient mathematical reasoning, math tutoring, automated assessment, lightweight reasoning agents, edge deployment, mobile applications, and latency-sensitive local inference
Context
66K
Reasoning
8/10
Speed
9/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; available through Microsoft’s Hugging Face repository and documented Azure AI Foundry availability
Efficient local or cloud text generation, multilingual applications, mathematics, coding, reasoning, retrieval-augmented generation, and edge deployment
Context
131K
Reasoning
7/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Generally available
Input
$0.00075 per 1,000 tokens in Microsoft's February 2025 Azure announcement; current Foundry pricing may vary
Output
$0.0003 per 1,000 tokens in Microsoft's February 2025 Azure announcement; current Foundry pricing may vary
Mathematical reasoning, educational tutoring, formal proof assistance, symbolic computation, local inference, edge deployment, and latency-sensitive applications
Context
128K
Reasoning
8/10
Speed
9/10
Status
Preview; currently available through Microsoft Foundry and open-weight distribution
Compact multimodal assistants, OCR, document and chart analysis, image question answering, speech recognition, speech translation, audio summarization, and private or local deployment
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Audio input
Multimodal input
Fine-tuning
Streaming
Status
Available; open-weight; listed in Microsoft Foundry and Hugging Face
Input
Not publicly listed as a current numeric price; Azure pricing page displays $-
Output
Not publicly listed as a current numeric price; Azure pricing page displays $-
Mathematical reasoning, coding, science, logic, algorithmic problem solving, research, and resource-constrained local deployments
Context
33K
Reasoning
8/10
Speed
7/10
Status
Preview; open-weight and currently listed in Microsoft Foundry
Input
No official per-token price published; open-weight model for self-hosted or separately priced managed deployment
Output
No official per-token price published; open-weight model for self-hosted or separately priced managed deployment
Mathematical reasoning, scientific problem solving, coding assistance, algorithmic tasks, local deployment, and reasoning-model research
Context
33K
Reasoning
8/10
Speed
5/10
Status
Current downloadable open-weight model
Short text-to-video and image-to-video clips, cinematic experiments, advertising concepts, social content, and scenes with complex motion.
Reasoning
1/10
Speed
5/10
Capabilities
Image input
Multimodal input
Status
Legacy or superseded; no longer prominent in MiniMax's current first-party model catalog
Output
Historical launch configurations were priced per generated video rather than by tokens; documented configurations included 512p/6s, 512p/10s, 768p/6s, 768p/10s, and 1080p/6s.
Short cinematic videos, image animation, realistic human motion, stylized scenes, visual effects, advertising concepts, and social-media content.
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Legacy or superseded video model; still referenced in MiniMax consumer and platform offerings, while MiniMax H3 is the current primary video model in developer documentation.
Output
$0.28 per 768P/6s clip; $0.56 per 768P/10s clip; $0.49 per 1080P/6s clip
Fast image-to-video generation, high-volume short-form content, social media clips, advertisements, and rapid creative iteration
Reasoning
1/10
Speed
9/10
Capabilities
Image input
Multimodal input
Status
Legacy; current availability should be verified
Input
$0.19 per 768p 6-second clip; $0.32 per 768p 10-second clip; $0.33 per 1080p 6-second clip
Short text-driven video concepts, storyboards, and early Hailuo-style cinematic experiments
Reasoning
2/10
Speed
6/10
Status
Legacy and superseded; no longer listed in MiniMax's current primary video-generation catalog
Prompt-based image generation, reference-guided variations, commercial visuals, and batch creative production
Capabilities
Image input
Multimodal input
Status
Current and accessible through MiniMax's image-generation console and API platform
Multilingual audio transcription, meeting transcription, speaker-labeled transcripts, live captions, call analysis, and subtitle generation
Reasoning
1/10
Speed
7/10
Capabilities
Audio input
Streaming
Status
Current public model
Input
$0.38 per hour of processed audio
Output
No separate output charge; billing is based on input audio duration
Fast short-form text-to-video, image-to-video, reference-guided generation, and synchronized-audio production
Reasoning
2/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Current; commercially hosted by fal
Output
$0.05 per second at 480p; $0.08 per second at 768p; $0.16 per second at 1080p
Multimodal commercial video generation, reference-based editing, product and advertising content, short cinematic clips, and locally deployed 768p workflows
Reasoning
2/10
Speed
5/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Fine-tuning
Status
Current; open-weight release and hosted API available
Input
$0.08 per second for 768p video output; $0.13 per second for 2K video output; H3-Context-IR: $0.90 per million input tokens
Output
$0.08 per second for 768p video output; $0.13 per second for 2K video output; H3-Regenerate-2K: $0.05 per second of regenerated output; H3-Context-IR: $3.60 per million output tokens
Coding agents, multi-step tool workflows, long-context codebase analysis, research automation, and self-hosted experimentation
Context
197K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Legacy/open-weight model; current hosted availability and pricing are not confirmed in MiniMax's latest public model catalog
Input
$0.30 per 1 million input tokens at launch; current pricing unverified
Output
$1.20 per 1 million output tokens at launch; current pricing unverified
Multilingual software engineering, coding agents, tool-using workflows, application development, and cost-sensitive automation
Context
205K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Legacy but currently available
Input
$0.30 per 1 million tokens; prompt-cache read $0.03 per 1 million tokens; prompt-cache write $0.375 per 1 million tokens
Output
$1.20 per 1 million tokens
Coding agents, software engineering, search and browser agents, tool-calling workflows, office automation, and long-context technical work
Context
205K
Reasoning
9/10
Speed
9/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Current and accessible; open-weight model; also available through the MiniMax API and MiniMax Agent
Input
$0.15 per 1 million input tokens for standard-speed M2.5 at launch; current prices may vary by platform and endpoint
Output
$1.20 per 1 million output tokens for standard-speed M2.5
Agentic software engineering, repository-level coding, production debugging, complex tool workflows, office document automation, and long-context professional tasks
Context
197K
Reasoning
8/10
Speed
8/10
Status
Current; available through MiniMax API, MiniMax Agent, and downloadable open weights
Input
$0.30 per 1 million tokens; cache read $0.06 per 1 million tokens; cache write $0.375 per 1 million tokens
Output
$1.20 per 1 million tokens
Low-latency coding assistants, multilingual software development, tool-using agents, long-horizon workflows, and interactive office automation
Context
205K
Reasoning
8/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Current and available through the MiniMax Open Platform API; historical model variant
Input
Â¥4.20 per 1M input tokens
Output
Â¥16.80 per 1M output tokens
Low-latency coding assistants, software-engineering agents, tool-using workflows, search tasks, and long-context productivity automation
Context
205K
Reasoning
8/10
Speed
10/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Current and available; highspeed variant of MiniMax M2.5
Input
$0.30 per 1 million tokens
Output
$2.40 per 1 million tokens
Low-latency coding assistants, software-engineering agents, tool-calling workflows, and interactive developer applications
Context
205K
Reasoning
8/10
Speed
10/10
Capabilities
Tool use
Streaming
Status
Current and available through the MiniMax API Platform; highspeed variant of MiniMax M2.7
Input
$0.60 per million tokens; cache read $0.06 per million tokens; cache write $0.375 per million tokens
Output
$2.40 per million tokens
Long-context coding agents, autonomous tool-using workflows, multimodal document and video analysis, and private deployment
Context
1M
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Active; open-weight and available through the MiniMax API
Input
$0.30 per million tokens for context up to 512K; $0.60 per million tokens for context from 512K to 1M. Listed rates are subject to MiniMax's current pricing terms.
Output
$1.20 per million tokens for context up to 512K; $2.40 per million tokens for context from 512K to 1M. Listed rates are subject to MiniMax's current pricing terms.
Generating complete songs with expressive vocals, lyrics, melodies, instrumental arrangements, duets, a cappella passages, and cinematic musical soundscapes.
Reasoning
1/10
Speed
6/10
Status
Legacy or limited availability; MiniMax states that music models were no longer available through Token Plan from 2026-08-20, but a full model retirement date was not verified.
Text-to-music, instrumental generation, game and video scoring, detailed musical direction, and genre reinterpretation with Cover mode
Capabilities
Audio input
Multimodal input
Status
Legacy or superseded; current public MiniMax navigation highlights Music 3.0, and Music 2.6 availability should be verified before use
Local generation of complete songs from lyrics and structured musical descriptions
Context
5K
Reasoning
1/10
Speed
4/10
Status
Open-weight model; MiniMax Token Plan access discontinued on 2026-08-20
Reinterpreting existing songs in new genres, vocal styles, arrangements, and production directions while preserving the source melody
Reasoning
1/10
Speed
6/10
Capabilities
Audio input
Multimodal input
Status
Current; specialized music-cover model introduced with MiniMax Music 2.6
Input
Not applicable to token pricing; audio-generation pricing varies by MiniMax platform or deployment
Output
Not applicable to token pricing; charged per generated audio result where applicable
High-quality multilingual voiceovers, audiobooks, narration, digital characters, advertising, education, and zero-shot voice cloning
Context
10K
Reasoning
1/10
Speed
6/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Legacy or older generation; still accessible through some partner platforms, but not listed as a core model on MiniMax's current global pricing page
Input
CNY 3.5 per 10,000 characters on Alibaba Cloud Model Studio in China; current direct MiniMax pricing for this exact model is not verified
Low-latency multilingual text-to-speech, streaming voice agents, interactive applications, expressive narration, and voice cloning
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Legacy or superseded; exact current first-party availability is unclear
Real-time voice agents, conversational assistants, customer-service automation, interactive characters, multilingual speech, and low-latency text-to-speech
Reasoning
1/10
Speed
9/10
Status
Legacy or superseded; current first-party availability is unverified
High-quality voiceovers, audiobooks, narration, localization, e-learning, game dialogue, accessibility audio, and production speech
Status
Legacy or transition-era model; not prominently listed in MiniMax's current first-party speech catalog as of September 25, 2026
High-quality expressive narration, audiobooks, podcasts, advertising, character voices, multilingual speech, and applications prioritizing audio fidelity over the lowest latency
Status
Current and available through the MiniMax API
Input
$100 per 1 million characters for text-to-audio usage
Real-time text-to-speech, voice assistants, conversational agents, interactive applications, multilingual narration, gaming characters and expressive voice experiences
Reasoning
1/10
Speed
9/10
Status
Current and accessible through the MiniMax Open Platform API
Input
$60 per 1 million characters
Text-to-video generation with explicit cinematic camera-movement direction, short advertising concepts, storyboards, and controlled visual experiments
Reasoning
1/10
Speed
6/10
Status
Legacy or limited availability; current official catalog status and continued first-party access are not clearly verified
Low-latency IDE autocomplete, fill-in-the-middle completion, code generation, code editing, test generation and developer assistants
Context
128K
Reasoning
6/10
Speed
9/10
Capabilities
Tool use
Streaming
Input
$0.30 per million tokens; cached input $0.03 per million tokens
Output
$0.90 per million tokens
Semantic code search, repository retrieval, coding-agent RAG, code similarity, duplicate detection, clustering, and code analytics
Input
$0.15 per million input tokens
Lean 4 theorem proving, formal verification, autoformalization, proof debugging, and agentic proof engineering
Context
256K
Reasoning
9/10
Speed
7/10
Status
Public Preview; scheduled for retirement on 2026-09-30
Low-cost edge and local inference, image-aware assistants, document analysis, structured extraction, lightweight agents, task routing, and privacy-sensitive deployments.
Context
256K
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Active; generally available
Input
$0.10 per 1 million tokens; cached input $0.01 per 1 million tokens
Output
$0.10 per 1 million tokens
Efficient edge and local inference, image understanding, document workflows, structured extraction, lightweight agents, and high-volume text generation.
Context
256K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Active; generally available
Input
$0.15 per million tokens; cached input $0.015 per million tokens
Output
$0.15 per million tokens
Private assistants, local vision-language applications, multilingual workloads, document and image analysis, and cost-efficient agentic systems
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Active; generally available
Input
$0.20 per 1 million tokens; cached input $0.02 per 1 million tokens
Output
$0.20 per 1 million tokens
Semantic search, retrieval-augmented generation, vector databases, document classification, clustering, duplicate detection and general text retrieval
Status
Generally available
Input
$0.10 per 1 million tokens
Output
$0.10 per 1 million tokens
Long-context enterprise assistants, multilingual applications, image-aware document analysis, agentic workflows, coding, RAG and self-hosted sovereign deployments
Context
256K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Active; generally available; open-weight
Input
$0.50 per 1M tokens; cached input $0.05 per 1M tokens
Output
$1.50 per 1M tokens
Agentic coding, software engineering, long-context analysis, multimodal document workflows, structured outputs and multi-step tool use
Context
256K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
GA; currently available; open weights
Input
$1.50 per million input tokens; $0.15 per million cached input tokens
Output
$7.50 per million output tokens
Text moderation, conversational safety classification, content filtering, policy enforcement, guardrails, and jailbreaking detection
Context
131K
Reasoning
2/10
Speed
8/10
Status
Active; generally available; Premier
High-volume OCR, structured document extraction, enterprise search, RAG ingestion, invoice processing, compliance workflows, and document automation
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Generally available; superseded by OCR 4.1 as Mistral's latest OCR model
Input
$4 per 1,000 pages; $0.40 per 1,000 cached pages; Batch API pricing reported at $2 per 1,000 pages
Output
Included in page-based processing price; no separate output-token price
Cost-efficient general chat, multimodal document analysis, coding, agentic workflows, and configurable reasoning
Context
256K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Active; generally available
Input
$0.15 per 1M tokens; cached input $0.015 per 1M tokens
Output
$0.60 per 1M tokens
High-volume document extraction, scanned forms, handwriting, invoices, complex tables, archival digitization, and document-to-knowledge pipelines.
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Legacy; available for existing integrations and production workloads. OCR 4 is the newer model.
Input
$2 per 1,000 pages
Output
$3 per 1,000 annotated pages
OCR, document parsing, structured extraction, enterprise search, RAG ingestion, invoice processing, and document AI workflows
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Generally Available
Input
$4 per 1,000 pages; cached input $0.40 per 1,000 pages
Policy-based moderation of prompts, model responses, refusals, text, images, and text-image combinations
Context
33K
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Streaming
Status
Public Preview; open weights; self-hosted
Batch transcription of meetings, interviews, calls, subtitles, compliance recordings, and searchable audio
Reasoning
1/10
Speed
8/10
Status
Active; GA; Premier hosted model
Live speech transcription, realtime captions, voice interfaces, realtime note-taking and speech-to-speech pipelines
Reasoning
1/10
Speed
10/10
Capabilities
Audio input
Streaming
Input
$0.006 per minute
Output
$0 per minute; transcription output is included in the audio-minute price
Production-scale audio understanding, multilingual transcription, audio Q&A, meeting and call summarization, speech translation, and voice-driven function calling
Context
32K
Reasoning
6/10
Speed
6/10
Capabilities
Audio input
Multimodal input
Tool use
Fine-tuning
Status
Active; generally available
Input
$0.004 per minute of audio; $0.10 per 1 million input tokens
Output
$0.40 per 1 million output tokens
Multilingual voice generation, expressive voice agents, zero-shot voice cloning, custom voice adaptation, and low-latency speech output
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Streaming
Status
GA; currently available through the Mistral API and Mistral Studio
Input
$0 per 1 million input characters
Output
$16 per 1 million output characters
Fine-tuning and research on speech recognition, audio understanding, audio classification, audio question answering, and speech-audio generation
Context
8K
Reasoning
4/10
Speed
5/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Streaming
Status
Current open-weight base model; not instruction-tuned
Input
No official hosted API pricing; downloadable open weights
Output
No official hosted API pricing; downloadable open weights
Self-hosted speech recognition, audio understanding, audio question answering, audio captioning, and spoken conversational agents
Reasoning
6/10
Speed
5/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Available as an open-weight self-hosted model
Input
No official hosted API pricing
Output
No official hosted API pricing
Repository-level software issue resolution, code repair, test generation, coding agents, and self-hosted software engineering workflows
Context
131K
Reasoning
7/10
Speed
4/10
Status
Available open-weight model
Self-hosted image and video understanding, OCR, long-document analysis, visual question answering, screenshot perception, and efficient multimodal applications
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Streaming
Status
Current open-weight/downloadable model; a newer Kimi-VL-A3B-Thinking-2506 variant is recommended for stronger multimodal reasoning
Input
No official Moonshot-hosted API price published; self-hosted/open-weight model
Output
No official Moonshot-hosted API price published; self-hosted/open-weight model
Local multimodal reasoning, mathematical visual question answering, OCR, document understanding, image and video analysis, and research on open-weight vision-language models
Context
131K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Available as downloadable open weights; superseded by Kimi-VL-A3B-Thinking-2506
Open-weight image and video reasoning, OCR, chart interpretation, visual mathematics, long PDFs, high-resolution screenshots and GUI-agent grounding
Context
131K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Current open-weight model; publicly available for self-hosted and third-party inference
Input
No official hosted API price; open-weight model
Output
No official hosted API price; open-weight model
Fine-tuning, foundation-model research, custom language systems, coding experiments, and self-hosted inference
Context
131K
Reasoning
8/10
Speed
3/10
Capabilities
Fine-tuning
Streaming
Status
Open-weight and downloadable; legacy relative to newer Kimi K2.x models but still accessible from the official Hugging Face repository
Input
No official hosted API price for the Base checkpoint; self-hosted weights
Output
No official hosted API price for the Base checkpoint; self-hosted weights
Open-weight coding assistants, tool-using agents, general-purpose chat, and self-hosted research deployments
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Open-weight checkpoint available; legacy relative to Moonshot AI's current hosted API catalog
Input
USD 0.60 per 1 million input tokens historically for the Kimi K2 preview API; current first-party pricing for this exact model is unverified
Output
USD 2.50 per 1 million output tokens historically for the Kimi K2 preview API; current first-party pricing for this exact model is unverified
Self-hosted reasoning agents, autonomous research, long-horizon tool workflows, coding, and complex multi-step analysis
Context
262K
Reasoning
9/10
Speed
5/10
Capabilities
Tool use
Streaming
Status
Retired from Moonshot's direct API; open-weight checkpoint remains available for self-hosted or third-party deployment
Long-horizon software engineering, agentic coding, visual document understanding, tool-using workflows, and multi-agent orchestration
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Status
Current; open-source model, available through Kimi, Kimi API, Kimi Code, and downloadable model weights
Long-horizon software engineering, repository-level coding, multi-file refactoring, debugging, coding agents, and tool-driven development workflows
Context
262K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and accessible through Kimi API; Kimi Code default service has moved to Kimi K2.8 Preview, while Kimi K2.7 Code remains available through API and Kimi K2.7 Code HighSpeed service.
Input
¥6.50 per 1M uncached input tokens; ¥1.30 per 1M cache-hit input tokens
Output
Â¥27.00 per 1M output tokens
Long-context software development, repository analysis, code completion, multi-file refactoring, and agentic coding workflows
Context
1.05M
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Preview; fully rolled out in Kimi Code
Input
Not publicly listed; accessed through Kimi Code membership and quota
Output
Not publicly listed; accessed through Kimi Code membership and quota
Multimodal coding, visual debugging, long-context analysis, tool-using agents, and complex research or office workflows
Context
256K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Input
$0.60 per 1 million tokens; cached input approximately $0.10 per 1 million tokens
Output
$3.00 per 1 million tokens
Long-context coding, software engineering, multimodal document and video understanding, agentic workflows, technical research, and complex reasoning
Context
1M
Reasoning
9/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current and available; open-weight model
Input
$3.00 per MTok cache-miss input; $0.30 per MTok cache-hit input
Output
$15.00 per MTok
Long-context text generation, local inference, continued pretraining, research, and fine-tuning
Context
1.05M
Reasoning
7/10
Speed
9/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; publicly downloadable
Input
Not applicable; open-weight checkpoint with no official hosted API price
Output
Not applicable; open-weight checkpoint with no official hosted API price
Long-context text generation, self-hosted assistants, document analysis, research, and efficient local inference
Context
1.05M
Reasoning
7/10
Speed
9/10
Status
Current open-weight instruction-tuned checkpoint
Self-hosted text generation, language-model research, efficient MoE inference, code and mathematics experimentation, and fine-tuning research
Context
8K
Reasoning
6/10
Speed
8/10
Status
Available open-weight pretrained checkpoint
Self-hosted instruction following, general text generation, research, experimentation, and cost-conscious local inference
Context
8K
Reasoning
6/10
Speed
8/10
Status
Available open-weight checkpoint
Native-resolution image feature extraction, vision-language model backbones, high-resolution document and image understanding, and multimodal research
Status
Current open-weight model; available for local use through Hugging Face Transformers; not deployed by a Hugging Face Inference Provider
General-purpose semantic search, retrieval, document similarity, clustering, and text classification
Context
500
Reasoning
1/10
Speed
8/10
Status
Current and accessible through CLOVA Studio's Embedding API
Sentence similarity, semantic search, document relatedness, clustering, and text classification features
Context
500
Reasoning
1/10
Speed
8/10
Status
Current and available through the CLOVA Studio Embedding API
Fast, cost-sensitive Korean text generation, classification, summarization, report drafting, data expansion and customized enterprise chatbots
Context
4K
Reasoning
4/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current and available through CLOVA Studio and related NAVER Cloud services
Input
Current standard inference price not clearly exposed in the retrieved official pricing table; NAVER's launch announcement described HCX-DASH-001 as costing approximately one-fifth of HCX-003.
Output
Current standard inference price not clearly exposed in the retrieved official pricing table; NAVER's launch announcement described HCX-DASH-001 as costing approximately one-fifth of HCX-003.
Korean-focused text generation, summarization, extraction, classification, tuning, and batch text workflows
Context
8K
Reasoning
5/10
Speed
7/10
Capabilities
Fine-tuning
Streaming
Status
Current and available through CLOVA Studio
Korean-language business applications, image understanding, document and visual analysis, instruction following, and API-based assistants.
Context
128K
Reasoning
6/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Input
Priced per 1,000 input tokens; current amount not displayed in the accessible official pricing table.
Output
Priced per 1,000 output tokens; current amount not displayed in the accessible official pricing table.
Complex reasoning, mathematics, science, language reasoning, writing, long-context text generation, and Korean-language enterprise applications
Context
128K
Reasoning
8/10
Speed
5/10
Capabilities
Tool use
Streaming
Status
Current and accessible through CLOVA Studio
Fast, high-throughput text generation, classification, summarization, simple extraction, and function-calling workflows
Context
32K
Reasoning
4/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Visual encoding for HyperCLOVA X SEED 4B and edge-oriented multimodal applications
Reasoning
0/10
Speed
0/10
Capabilities
Image input
Multimodal input
Status
Current; proprietary vision encoder component for HyperCLOVA X SEED 4B
Lightweight Korean conversational interfaces, mobile and edge applications, smart-home devices, wearables, and customer-support chatbots
Context
4K
Reasoning
3/10
Speed
9/10
Status
Current open-weight model; available for download
Korean-language applications, lightweight local inference, basic translation, education, business communication, specialized chatbots, and domain fine-tuning.
Context
16K
Reasoning
4/10
Speed
8/10
Status
Current downloadable open-weight model; Hugging Face repository is gated and requires acceptance of access conditions.
Korean image and video understanding, visual question answering, chart and diagram interpretation, OCR-assisted analysis, tourism and cultural applications, and locally deployable fine-tuned systems.
Context
16K
Reasoning
4/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API price identified; model weights are available for download under the hyperclovax-seed license.
Output
No official hosted API price identified; deployment costs depend on infrastructure or hosting provider.
Korean visual and document understanding, video-and-audio analysis, edge AI, public-sector systems, defense intelligence, and air-gapped deployments
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Publicly announced; current model identity, with detailed commercial/API availability not publicly specified
Korean-first any-to-any multimodal assistants, speech and vision applications, multimodal research, and self-hosted deployments
Context
33K
Reasoning
7/10
Speed
6/10
Outputs
Text
Image
Speech
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current open-weight model
Korean-language reasoning, mathematics, coding, instruction following, tool-connected agents, and self-hosted commercial applications
Context
33K
Reasoning
7/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Current open-weight model; free for commercial use under the HyperCLOVA X SEED license
Input
Free model weights; no official hosted API input price published for this exact checkpoint
Output
Free model weights; no official hosted API output price published for this exact checkpoint
Korean-language reasoning, visual question answering, long-context multimodal analysis, document and chart understanding, coding assistance, and tool-using AI agents
Context
128K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current; open-weight/open-source release
Input
No official hosted API price published; self-hosting costs depend on infrastructure
Output
No official hosted API price published; self-hosting costs depend on infrastructure
Real-time or batch speaker identification and active-speaker tagging in broadcast, video communication, dubbing, localization, conferencing, and media analytics workflows.
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Video input
Multimodal input
Streaming
Status
Current; downloadable NVIDIA NIM endpoint
Real-time video relighting, virtual production, media effects, HDR-based lighting changes, and foreground/background compositing.
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Current and accessible through NVIDIA AI for Media and NVIDIA NIM documentation; version 1.1.0 is documented.
Autonomous-driving research, trajectory prediction, interpretable motion planning, navigation-conditioned driving, visual question answering and safety-oriented model evaluation
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current; open weights available; also available through NVIDIA Alpamayo 1.5 NIM
Multilingual offline speech transcription, speech translation, and NeMo-based ASR research or deployment
Reasoning
1/10
Speed
6/10
Status
Active; open-weight checkpoint and NVIDIA deployment options available
Fast English speech-to-text transcription, NeMo experimentation, domain fine-tuning, and Riva-based ASR deployment
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Fine-tuning
Streaming
Status
Available downloadable checkpoint; older NeMo ASR model
Controllable video world generation, robotics sim-to-real augmentation, autonomous-vehicle simulation, and Physical AI synthetic-data generation
Reasoning
1/10
Speed
3/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Available; legacy relative to Cosmos 3; original repository under limited maintenance
Input
No official token-based hosted price; NVIDIA lists a free NIM endpoint. Downloadable model is intended for self-hosted deployment.
Output
No official token-based hosted price; video-generation costs depend on deployment infrastructure and hardware.
Edge physical AI, robotics, visual reasoning, world simulation, video generation, and action-policy prototyping
Context
131K
Reasoning
7/10
Speed
9/10
Outputs
Text
Image
Video
Actions
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current; open model; gated Hugging Face access
Physical AI, robotics, autonomous-vehicle simulation, multimodal world generation, future-state prediction, action reasoning, and synthetic training data
Reasoning
7/10
Speed
6/10
Outputs
Text
Image
Video
Audio
Actions
Capabilities
Image input
Audio input
Video input
Multimodal input
Fine-tuning
Status
Current; downloadable open-weight model and available through an NVIDIA NIM endpoint
High-quality Physical AI simulation, synthetic-data generation, robotics and autonomous-vehicle research, multimodal world modeling, and teacher-model distillation
Context
262K
Reasoning
8/10
Speed
2/10
Outputs
Text
Image
Video
Audio
Actions
Capabilities
Image input
Audio input
Video input
Multimodal input
Fine-tuning
Status
Current; open-weight model; commercially and non-commercially usable
Physical-world video and image understanding, robotic perception, embodied-agent planning, spatial-temporal reasoning, and Physical AI research
Context
256K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current; downloadable and hosted NIM endpoint
Input
Free downloadable endpoint; hosted pricing not specified in the reviewed NVIDIA sources
Output
Free downloadable endpoint; hosted pricing not specified in the reviewed NVIDIA sources
Rapid global weather forecasting, ensemble simulation, climate-risk analysis, renewable-energy forecasting, and scientific weather-model research
Reasoning
0/10
Speed
9/10
Status
Current and downloadable; available through NVIDIA Earth-2 FourCastNet NIM and related deployment tooling
Input
Not publicly listed as a token-priced model; downloadable model and NIM access are subject to NVIDIA access and licensing terms
Output
Not publicly listed as a token-priced model; deployment and hosted inference pricing may depend on the selected NVIDIA service or infrastructure
De novo molecular design, fragment-constrained generation, linker design, scaffold decoration, hit generation, and lead optimization
Context
512
Reasoning
2/10
Speed
7/10
Status
Current; downloadable model and available as an NVIDIA NIM
Input
No public per-token model price found; downloadable weights are available under the NVIDIA Open Model License. NIM access is subject to NVIDIA API or deployment terms.
Output
No public per-token model price found; NIM output pricing is not specified in the reviewed official documentation.
Humanoid robot manipulation, cross-embodiment policy learning, robot demonstration fine-tuning, physical AI research, and action-sequence deployment
Reasoning
4/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current; general availability
Quantum-computing calibration plot interpretation, QPU bring-up and retuning workflows, experiment diagnosis, parameter extraction, fit-quality assessment, and domain-specific calibration agents.
Context
262K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Streaming
Status
Current; available through NVIDIA Build, NVIDIA NIM, and downloadable checkpoints
Generative lip dubbing, multilingual video localization, broadcasting, conferencing, and digital-human facial animation
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Current; downloadable model and NVIDIA LipSync NIM; private access may be required for some workflows
Multimodal semantic search, visual document retrieval, question-answer retrieval, vector databases, and retrieval-augmented generation
Context
10K
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and downloadable
Reranking text, document images, and image-text candidates in visual search, multimodal RAG, and question-answering retrieval pipelines
Context
8K
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Current; downloadable and available through NVIDIA NIM and retrieval APIs
Multilingual voice agents, accessibility, narration, audiobooks, dubbing, localization and interactive speech applications
Reasoning
1/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Low-latency multilingual text translation, speech-translation pipelines, and self-hosted NVIDIA GPU deployments
Reasoning
1/10
Speed
8/10
Status
Current and downloadable; available through NVIDIA NIM and NVIDIA Riva
Small-molecule generation, molecular embeddings, chemical-space exploration, lead optimization, and oracle-guided drug-design workflows
Status
Available through NVIDIA BioNeMo Framework and NVIDIA NIM; research and development model
Detecting table cells, rows, columns, and merged-cell structure in document images for OCR alignment, table reconstruction, document ingestion, and retrieval systems
Reasoning
1/10
Speed
7/10
Status
Current; downloadable and available through NVIDIA NIM/Build
Agentic reasoning, long-context analysis, coding, tool-calling workflows, retrieval-augmented generation, collaborative agents, and high-volume enterprise inference.
Context
1M
Reasoning
9/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current; open-weight; available through NVIDIA NIM, NVIDIA's hosted API trial endpoint, downloadable checkpoints, and self-hosted deployments.
Frontier reasoning, complex coding, long-context analysis, enterprise RAG, tool-using agents, and multi-agent workflows
Context
1M
Reasoning
10/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current; open-weight model with BF16 and NVFP4 checkpoints
Multilingual semantic search, dense retrieval, RAG, agentic retrieval, code search, and vector-based document matching
Status
Current; available through NVIDIA NIM and downloadable Hugging Face weights
Multimodal document intelligence, OCR, long-video and audio understanding, cross-modal reasoning, voice agents, and self-hosted enterprise inference
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; generally available open-weight checkpoint with hosted NVIDIA NIM access
Real-time full-duplex voice agents, interruptible conversational interfaces, speech-to-speech research, and NVIDIA GPU-based enterprise voice applications
Reasoning
5/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Streaming
Status
Early access; available for evaluation through NVIDIA NIM and qualified access programs
Input
Free endpoint for NVIDIA NIM trial access; no general production price published
Output
Free endpoint for NVIDIA NIM trial access; no general production price published
High-volume agent execution, long-running AI agents, reasoning, coding assistance, RAG, chat, low-latency text generation, and domain customization
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; NIM container available for early-access evaluation
Low-latency live speech-to-text, voice interfaces, live captions, and continuous audio streams
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Current and accessible through NVIDIA Speech NIM; streaming-only deployment
Multilingual text-and-image moderation, LLM and VLM guardrails, response safety evaluation, and custom enterprise safety policies
Context
128K
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current; open-weight model and downloadable NVIDIA NIM; latest NGC container version bf16-v1.1 as of September 17, 2026
Nemotron Graphic Elements
Detecting and localizing chart titles, axis labels, legends, mark labels, and value labels in document images
Reasoning
1/10
Speed
7/10
Status
Current and downloadable
English OCR, document ingestion, layout-aware text extraction, multimodal retrieval, RAG preprocessing, and enterprise document intelligence
Reasoning
1/10
Speed
8/10
Status
Available; English-only OCR model; newer Nemotron OCR v2 is available for updated English and multilingual OCR deployments
Multilingual OCR, scanned documents, forms, reports, charts, tables, image-based search, document ingestion, and retrieval-augmented generation preprocessing
Reasoning
1/10
Speed
9/10
Status
Current; downloadable model and available through NVIDIA NIM and NVIDIA-hosted services
Document page-layout detection before OCR, table extraction, indexing, enterprise document processing, and multimodal RAG pipelines
Reasoning
1/10
Speed
7/10
Status
Current; downloadable and available through NVIDIA NIM
Document OCR, layout analysis, table and chart extraction, retrieval pipelines, document indexing, and multimodal data curation
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Quantum-computing calibration plot analysis, experiment interpretation, fit-quality assessment, parameter extraction, and technical research workflows
Context
262K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current; downloadable open-weight model
NVIDIA Maxine Eye Contact
Gaze correction in video conferencing, telepresence, digital-human applications, and video-processing pipelines
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Status
Current; downloadable NVIDIA NIM model
English speech-to-text transcription, local ASR, offline processing, RAG audio extraction, and customized NeMo deployments
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Fine-tuning
Streaming
Status
Available open-weight checkpoint
Local English speech transcription, offline ASR, long-form audio processing, and custom ASR fine-tuning
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Fine-tuning
Status
Current and accessible open-weight model
Taiwanese Mandarin speech-to-text, Mandarin-English code-switching, live captions, voice interfaces, and NVIDIA GPU-hosted streaming or offline transcription.
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Streaming
Status
Listed in current NVIDIA Speech NIM documentation and model catalog; the associated NGC container is marked no longer supported.
Mandarin-English automatic speech recognition, code-switched transcription, streaming transcription, and self-hosted enterprise speech-to-text
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Streaming
Status
Current and downloadable; available through NVIDIA NIM and Riva interfaces
Input
No public per-token price; downloadable model and NVIDIA service terms apply
Output
No public per-token price; downloadable model and NVIDIA service terms apply
High-speed English speech transcription, captions, meeting transcription, audio search, and timestamped media workflows
Reasoning
1/10
Speed
10/10
Capabilities
Audio input
Fine-tuning
Streaming
Status
Current open-weight model; publicly available through Hugging Face and NVIDIA NeMo
Multilingual sentence and document translation, localization, marketing content, and developer translation workflows
Context
8K
Reasoning
2/10
Speed
7/10
Status
Current and accessible through NVIDIA NIM and downloadable model weights
Input
Free hosted endpoint listed on Build.NVIDIA.com; no self-hosted per-token price specified
Output
Free hosted endpoint listed on Build.NVIDIA.com; no self-hosted per-token price specified
Local or self-hosted multilingual sentence and document translation across English and 36 non-English languages
Context
8K
Reasoning
3/10
Speed
8/10
Status
Current and publicly downloadable; supported by NVIDIA NIM
Camera-only multi-view 3D perception, autonomous-driving scene analysis, bird's-eye-view visualization, and object tracking
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Current NVIDIA NIM endpoint; free endpoint access requires an NVIDIA API key
Real-time enhancement of speech captured with low-quality microphones in noisy or reverberant environments, including broadcast, conferencing, telecommunications, and media production.
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Streaming
Status
Current and available through NVIDIA NIM, hosted preview services, and NVIDIA audio software; downloadable deployment may require applicable NVIDIA licensing or subscription access.
AI-generated video detection, media authentication, digital forensics, content moderation, and media-integrity monitoring
Reasoning
1/10
Speed
8/10
Capabilities
Video input
Streaming
Status
Current; downloadable NIM and managed trial endpoint, with self-hosted deployment requiring AI for Media Private Access
Input
Free managed trial endpoint; self-hosted and enterprise licensing terms may apply
Professional video upscaling, broadcast enhancement, streaming pipelines, pre-encoding optimization, denoising, and deblurring
Reasoning
1/10
Speed
8/10
Capabilities
Video input
Multimodal input
Fine-tuning
Streaming
Status
Current; downloadable NVIDIA NIM
ChatGPT-style instant responses, general-purpose writing and analysis, image-aware conversations, and tool-assisted workflows
Context
400K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current rolling alias; underlying model snapshot is regularly updated
Input
$5.00 per 1 million input tokens; cached input $0.50 per 1 million tokens
Output
$30.00 per 1 million output tokens
Zero-shot image classification, image-text similarity, semantic image retrieval, multimodal indexing, and computer-vision research
Context
77
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Public research release with downloadable weights; not verified as a current OpenAI hosted API model
Codex CLI coding workflows, code question answering, code editing, repository tasks, and low-latency software-engineering assistance
Context
200K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access ended on 2026-02-12
Input
$1.50 per 1M input tokens; $0.375 per 1M cached input tokens
Output
$6.00 per 1M output tokens
Controlled browser automation, computer-use research, UI testing, and repetitive interface workflows
Context
8K
Reasoning
6/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Input
$3.00 per 1 million input tokens; tool-specific computer-use calls may incur separate fees
Output
$12.00 per 1 million output tokens
Historical research on text-to-image generation, legacy image workflows, and comparisons with newer OpenAI image models.
Capabilities
Image input
Multimodal input
Status
Retired; deprecated and removed from the OpenAI API on May 12, 2026.
Historical text-to-image generation, concept art, illustration, visual ideation, marketing imagery, and prompt-following research
Reasoning
1/10
Speed
6/10
Status
Retired; deprecated and removed from the OpenAI API on May 12, 2026
Input
Not applicable to current use; historical pricing was charged per generated image rather than per input token
Output
Historical API pricing started at $0.04 per 1024×1024 standard-quality image; higher prices applied to HD and larger formats
Maintaining legacy text-completion applications, historical GPT-3 base-model behavior, and existing compatible fine-tuned workflows before shutdown
Reasoning
2/10
Speed
7/10
Status
Deprecated; currently accessible but scheduled to shut down on 2026-09-28
Input
$0.40 per 1 million input tokens
Output
$0.40 per 1 million output tokens
Legacy text completion, code continuation, and inference from existing davinci-002 fine-tuned models before shutdown
Reasoning
3/10
Speed
6/10
Status
Deprecated; API access scheduled to shut down on September 28, 2026
Input
$2.00 per 1M tokens
Output
$2.00 per 1M tokens
Low-cost, high-volume text generation, summarization, classification, extraction, simple chatbots, and legacy API integrations
Context
16K
Reasoning
4/10
Speed
8/10
Status
Deprecated; still available through the OpenAI API
Input
$0.50 per 1 million input tokens
Output
$1.50 per 1 million output tokens
Maintaining established GPT-4 integrations, general-purpose text generation, analysis, writing, and coding workloads
Context
8K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Legacy; older high-intelligence GPT model
Input
$30 per 1 million prompt tokens
Output
$60 per 1 million completion tokens
Legacy high-context text and image analysis, function calling, JSON-mode workflows, and existing GPT-4 Turbo integrations
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Deprecated but currently accessible; scheduled for shutdown on October 23, 2026
Input
$10 per 1 million input tokens
Output
$30 per 1 million output tokens
Historical long-context text generation, document analysis, structured text generation, and general-purpose assistant applications.
Context
128K
Reasoning
7/10
Speed
7/10
Status
Retired; the gpt-4-turbo-preview alias pointed to gpt-4-0125-preview, which was shut down on 2026-03-26.
Input
$10 per 1 million tokens
Output
$30 per 1 million tokens
Software engineering, long-context document analysis, precise instruction following, structured extraction, tool-enabled agents, and image understanding
Context
1.05M
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Current; default GPT-4.1 alias with gpt-4.1-2025-04-14 snapshot
Input
$2.00 per 1M input tokens; $0.50 per 1M cached input tokens
Output
$8.00 per 1M output tokens
Fast, cost-efficient instruction following, coding assistance, image understanding, structured extraction, tool calling, and long-context API applications
Context
1.05M
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Input
$0.40 per 1 million tokens; cached input $0.10 per 1 million tokens
Output
$1.60 per 1 million tokens
High-volume, latency-sensitive classification, extraction, routing, summarization, lightweight assistants, image-assisted analysis, and simple tool-calling workflows
Context
1.05M
Reasoning
4/10
Speed
10/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Deprecated; currently accessible as of September 23, 2026; scheduled for shutdown on October 23, 2026
Input
$0.10 per 1M input tokens; $0.025 per 1M cached input tokens
Output
$0.40 per 1M output tokens
Historical general-purpose writing, creative work, nuanced communication, image understanding, programming assistance, and applications needing function calling or structured outputs.
Context
128K
Reasoning
7/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired from the API on 2025-07-14; retired from ChatGPT in June 2026
Input
$75.00 per 1 million input tokens; $37.50 per 1 million cached input tokens
Output
$150.00 per 1 million output tokens
Fast general-purpose conversations, vision, voice interactions, coding, and everyday productivity
Context
128K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
General-purpose assistants, image understanding, coding help, structured extraction, multilingual generation, and latency-sensitive API workflows
Context
128K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Current in the OpenAI API; retired from ChatGPT on 2026-02-13. The gpt-4o-2024-05-13 snapshot is scheduled for API shutdown on 2026-10-23.
Input
$2.50 per 1M input tokens; $1.25 per 1M cached input tokens
Output
$10.00 per 1M output tokens
Voice assistants, spoken conversational agents, audio-enabled customer service, and applications requiring direct audio understanding and speech generation
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Retired; API access ended May 7, 2026
Input
$2.50 per 1M text input tokens; $40 per 1M audio input tokens
Output
$10.00 per 1M text output tokens; $80 per 1M audio output tokens
Low-cost, high-volume text and image understanding, classification, extraction, translation, tagging, customer support, routing, and structured data generation
Context
128K
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Fine-tuning
Streaming
Status
Current canonical model alias; dated snapshot gpt-4o-mini-2024-07-18 is available
Input
$0.15 per 1M input tokens
Output
$0.60 per 1M output tokens
Lower-cost audio understanding, conversational voice interfaces, and applications requiring text and spoken-audio input/output
Context
128K
Reasoning
3/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Deprecated; scheduled for API shutdown on 2027-01-20
Input
Text: $0.15 per 1M tokens; audio: $10.00 per 1M tokens
Output
Text: $0.60 per 1M tokens; audio: $20.00 per 1M tokens
Low-cost realtime voice assistants, speech-to-speech interfaces, interactive audio applications, and conversational prototypes
Context
16K
Reasoning
4/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Status
Deprecated; scheduled for API removal on 2027-01-20
Input
Text: $0.60 per 1M input tokens; audio: $10.00 per 1M audio tokens; cached input: $0.30 per 1M tokens
Output
Text: $2.40 per 1M output tokens; audio: $20.00 per 1M audio tokens
Low-latency voice assistants, speech-to-speech applications, live translation, language learning, and interactive customer support
Context
32K
Reasoning
6/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Status
Retired; API access ended 2026-05-07
Input
$5 per 1M text tokens; $40 per 1M audio tokens; cached text and audio input $2.50 per 1M tokens
Output
$20 per 1M text tokens; $80 per 1M audio tokens
Historical web-search applications built around OpenAI Chat Completions
Context
128K
Reasoning
5/10
Speed
6/10
Capabilities
Tool use
Streaming
Status
Retired; shut down on 2026-07-23
Input
$2.50 per 1 million input tokens; historical web-search tool fees applied separately per search call
Output
$10.00 per 1 million output tokens
Accurate speech-to-text conversion, meeting transcription, call transcription, voice-agent input, and prompted domain-specific transcription
Context
16K
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Deprecated; currently accessible; scheduled for API shutdown on 2027-02-26
Input
$2.50 per 1M audio input tokens
Output
$10.00 per 1M audio output tokens
Lower-cost multilingual speech transcription, meeting notes, call-center transcripts, voice-note conversion, and audio-to-text pipelines
Context
16K
Reasoning
1/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Current and available
Input
$1.25 per 1 million audio tokens
Output
$5.00 per 1 million audio tokens
Fast, controllable text-to-speech for narration, voice interfaces, customer service, accessibility, and realtime audio applications.
Context
2K
Reasoning
1/10
Speed
9/10
Status
Current; the canonical alias currently points to the gpt-4o-mini-tts-2025-12-15 snapshot.
Input
$0.60 per 1M text input tokens
Output
$12.00 per 1M audio output tokens
GPT-4o Mini Search Preview
Legacy Chat Completions applications requiring low-cost, search-grounded text responses
Context
128K
Reasoning
5/10
Speed
9/10
Capabilities
Tool use
Web search
Streaming
Status
Retired; access shut down on July 23, 2026
Input
$0.15 per 1 million input tokens, plus a separate fee per web-search tool call
Output
$0.60 per 1 million output tokens
Multi-speaker meeting, interview, call, podcast, and research transcription with speaker labels.
Context
16K
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Streaming
Status
Deprecated; currently accessible and scheduled for removal from the API on February 26, 2027.
Input
$2.50 per 1M audio tokens
Output
$10.00 per 1M audio tokens
Complex coding, reasoning, research, long-context analysis, visual understanding, tool-using agents, and structured professional workflows
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current canonical alias, but previous-generation model; dated snapshot gpt-5-2025-08-07 is deprecated and scheduled for API shutdown on 2026-12-11
Input
$1.25 per 1 million input tokens; cached input $0.125 per 1 million tokens
Output
$10.00 per 1 million output tokens
ChatGPT-aligned conversational applications, text generation, image-aware question answering, structured outputs, and tool-enabled workflows requiring GPT-5 compatibility
Context
128K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Input
$1.25 per 1M input tokens; $0.125 per 1M cached input tokens
Output
$10.00 per 1M output tokens
Cost-sensitive reasoning, coding assistance, structured extraction, document processing, high-volume automation, and tool-enabled workflows
Context
400K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current API alias; dated snapshot gpt-5-mini-2025-08-07 is deprecated
Input
US$0.25 per 1 million input tokens; cached input US$0.025 per 1 million tokens
Output
US$2.00 per 1 million output tokens
High-volume classification, summarization, extraction, ranking, routing, image-assisted analysis, and lightweight coding subagents
Context
400K
Reasoning
7/10
Speed
10/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Deprecated dated snapshot; currently accessible until scheduled shutdown on 2026-12-11
Input
$0.05 per 1 million tokens; cached input $0.005 per 1 million tokens
Output
$0.40 per 1 million tokens
Difficult research, mathematics, science, complex coding, high-stakes analysis, and tool-using workflows where maximum answer quality matters more than latency or cost.
Context
400K
Reasoning
10/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current canonical alias; dated snapshot gpt-5-pro-2025-10-06 is deprecated and scheduled for shutdown on 2026-12-11.
Input
$15 per 1M input tokens
Output
$120 per 1M output tokens
Agentic software engineering, repository-level coding, code review, debugging, refactoring, test generation, and frontend work using screenshots
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access shut down on 2026-07-23
Input
$1.25 per 1M input tokens; $0.125 per 1M cached input tokens
Output
$10.00 per 1M output tokens
Coding, long-context analysis, tool-using agents, structured outputs, and multi-step workflows
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$1.25 per 1 million input tokens; $0.125 per 1 million cached input tokens
Output
$10.00 per 1 million output tokens
Conversational assistants, instruction following, image-grounded chat, structured extraction, streaming responses, and tool-using API workflows.
Context
128K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; the API alias gpt-5.1-chat-latest was shut down on 2026-07-23. GPT-5.1 models were retired from ChatGPT on 2026-03-11.
Input
$1.25 per 1 million input tokens; cached input $0.125 per 1 million tokens
Output
$10.00 per 1 million output tokens
Agentic software engineering, code generation, debugging, refactoring, testing, code review, and long-running Codex workflows
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access shut down on July 23, 2026
Input
$1.25 per 1M input tokens; $0.125 per 1M cached input tokens
Output
$10.00 per 1M output tokens
Long-running agentic coding, repository-scale refactoring, multi-file implementation, debugging, code review, pull-request creation, and extended Codex workflows.
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access ended 2026-07-23
Input
$1.25 per 1M tokens; cached input $0.125 per 1M tokens
Output
$10.00 per 1M tokens
Cost-sensitive agentic coding, code editing, repository maintenance, and Codex-style workflows
Context
400K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access shut down on 2026-07-23
Input
$0.25 per 1 million tokens; cached input $0.025 per 1 million tokens
Output
$2.00 per 1 million tokens
Complex professional work, long-context analysis, coding, document and spreadsheet workflows, visual understanding, and multi-step agents
Context
400K
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Currently available; previous flagship model
Input
$1.75 per 1M input tokens; $0.175 per 1M cached input tokens
Output
$14.00 per 1M output tokens
ChatGPT-aligned conversational applications, general writing, summarization, translation, vision-enabled assistants, and tool-calling workflows
Context
128K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access ended on 2026-08-10
Input
$1.75 per 1 million input tokens; $0.175 per 1 million cached input tokens
Output
$14.00 per 1 million output tokens
Complex professional reasoning, advanced analysis, scientific and mathematical work, high-quality coding, long-context document analysis, and tool-using workflows.
Context
400K
Reasoning
10/10
Speed
4/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Previous Pro model; currently available through the Responses API
Input
$21 per 1M tokens
Output
$168 per 1M tokens
Long-horizon agentic coding, large refactors, code migrations, repository-scale changes, terminal workflows, Windows development and defensive cybersecurity
Context
400K
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access shut down on 2026-07-23
Input
$1.75 per 1 million input tokens
Output
$14.00 per 1 million output tokens
Fast general-purpose conversation, writing, summarization, text-and-image understanding, streaming responses, and function-calling applications
Context
128K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Retired; API access ended 2026-08-10
Input
$1.75 per 1M tokens; cached input $0.175 per 1M tokens
Output
$14.00 per 1M tokens
Long-running agentic software engineering, codebase maintenance, debugging, testing, web development, tool-driven development, and technical computer workflows
Context
400K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current and available through OpenAI API and Codex surfaces
Input
$1.75 per 1 million input tokens
Output
$14.00 per 1 million output tokens
Complex professional work, advanced reasoning, software engineering, long-horizon agents, visual document analysis, computer use, research, and tool-heavy workflows
Context
1.05M
Reasoning
10/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$2.50 per 1 million input tokens; $0.25 per 1 million cached input tokens
Output
$15.00 per 1 million output tokens
High-volume coding assistants, computer-use agents, subagents, tool calling, image reasoning, document workflows, and latency-sensitive applications
Context
400K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$0.75 per 1 million input tokens; $0.075 per 1 million cached input tokens
Output
$4.50 per 1 million output tokens
High-volume classification, data extraction, ranking, image understanding, routing, and lightweight coding subagents
Context
400K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; API-only model
Input
$0.20 per 1M input tokens; $0.02 per 1M cached input tokens
Output
$1.25 per 1M output tokens
High-stakes reasoning, professional knowledge work, long-context analysis, complex coding, web research and agentic workflows requiring maximum answer quality
Context
1.05M
Reasoning
10/10
Speed
4/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; available in ChatGPT for Pro and Enterprise users and in the Responses API for developers
Input
$30 per 1 million input tokens
Output
$180 per 1 million output tokens
Authorized vulnerability research, defensive cybersecurity operations, malware analysis, security testing, and binary reverse engineering.
Status
Deprecated; currently accessible through restricted Trusted Access for Cyber channels; scheduled for API shutdown on October 1, 2026.
Complex coding, long-context research, professional analysis, tool-heavy agents, computer use, and multi-step workflow execution
Context
1.05M
Reasoning
10/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; available through the OpenAI API, ChatGPT, and Codex
Input
$5.00 per 1 million input tokens; cached input $0.50 per 1 million tokens
Output
$30.00 per 1 million output tokens
High-accuracy reasoning, complex coding, long-context research, data analysis, and multi-step professional workflows
Context
1.05M
Reasoning
10/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Input
USD 30 per 1 million input tokens; no cached-input discount
Output
USD 180 per 1 million output tokens
Authorized vulnerability research, exploit validation, exploit-chain development, advanced security testing, vulnerability triage, and defensive cybersecurity agents
Context
400K
Reasoning
10/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; restricted access through OpenAI Daybreak Red with separate approval and provisioning
Input
$12.50 per 1 million input tokens; cached input $1.25 per 1 million tokens
Output
$75.00 per 1 million output tokens
High-volume classification, summarization, routing, extraction, document understanding, agent automation, routine coding assistance, and cost-sensitive tool-using applications.
Context
1.05M
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$0.20 per 1 million input tokens; cached input $0.02 per 1 million tokens; cache writes billed at 1.25x the uncached input rate. Requests with more than 272,000 input tokens are priced at 2x input for the full request.
Output
$1.20 per 1 million output tokens. Requests with more than 272,000 input tokens are priced at 1.5x output for the full request.
Complex reasoning, coding, research, cybersecurity, science, long-context analysis, document-heavy workflows, and tool-using agents
Context
1.05M
Reasoning
10/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Generally available
Input
$4 per 1 million input tokens; cached input $0.40 per 1 million tokens
Output
$20 per 1 million output tokens
Cost-conscious reasoning, coding agents, long-context analysis, structured business automation, research workflows, and tool-enabled production applications
Context
1.05M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Generally available
Input
$2.00 per 1 million input tokens; $0.20 per 1 million cached input tokens
Output
$12.00 per 1 million output tokens
Complex reasoning, agentic coding, computer use, web research, scientific and professional workflows, and long-context document tasks
Context
1.05M
Reasoning
10/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; rolling out through the OpenAI API and selected ChatGPT, Azure, and Amazon Bedrock offerings
Input
$10.00 per 1 million input tokens for Standard short-context processing; $1.00 per 1 million cached input tokens; $12.50 per 1 million cache-write tokens. Long-context input is $20.00 per 1 million tokens.
Output
$50.00 per 1 million output tokens for Standard short-context processing; $75.00 per 1 million output tokens for long-context processing. Batch and Flex processing are priced at 50% of Standard rates.
High-volume reasoning, document analysis, coding assistance, retrieval-augmented generation, and repeatable agent workflows
Context
1.05M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current and available
Input
$0.10 per 1 million input tokens; $0.01 cached input; $0.125 cache writes. Long-context rates are $0.20 input, $0.02 cached input, and $0.25 cache writes per 1 million tokens.
Output
$0.50 per 1 million output tokens for short context; $0.75 per 1 million output tokens for long context.
Complex coding, long-context reasoning, software engineering, research, computer use, and agentic workflows with tools.
Context
1.05M
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; generally available through the OpenAI API
Input
USD 2.00 per 1M input tokens for short context; USD 4.00 per 1M input tokens for long context. Cached input is USD 0.20 short-context or USD 0.40 long-context per 1M tokens. Cache writes are USD 2.50 short-context or USD 5.00 long-context per 1M tokens un
Output
USD 10.00 per 1M output tokens for short context; USD 15.00 per 1M output tokens for long context under Standard processing.
Audio-enabled chat applications, voice interfaces, spoken assistants, and applications requiring direct audio understanding and generation through Chat Completions.
Context
128K
Reasoning
5/10
Speed
7/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Deprecated; scheduled for shutdown on January 20, 2027
Input
$2.50 per 1M text tokens; $32.00 per 1M audio tokens
Output
$10.00 per 1M text tokens; $64.00 per 1M audio tokens
Audio-in, audio-out conversational applications using the Chat Completions API, including voice assistants and tool-enabled spoken interfaces.
Context
128K
Reasoning
5/10
Speed
7/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Current; generally available
Input
Text: $2.50 per 1M tokens; audio: $32.00 per 1M audio tokens
Output
Text: $10.00 per 1M tokens; audio: $64.00 per 1M audio tokens
Cost-sensitive, turn-based audio conversations, voice assistants, and audio-enabled applications using function calling
Context
128K
Reasoning
5/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Tool use
Status
Deprecated; currently accessible but scheduled for API removal on 2027-01-20
Input
$0.60 per 1 million text input tokens
Output
$2.40 per 1 million text output tokens
Precise image editing, detailed creative work, high-fidelity generation, infographics, layouts, and workflows where fewer retries matter more than minimum latency
Capabilities
Image input
Multimodal input
Input
Text input: $5.00 per 1M tokens; image input: $8.00 per 1M tokens; cached text input: $1.25 per 1M tokens; cached image input: $2.00 per 1M tokens
Output
Image output: $30.00 per 1M tokens
Natural low-latency voice agents, customer support, conversational workflows, live assistance, and applications requiring interruption-aware speech interaction
Reasoning
7/10
Speed
10/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Current; available in the OpenAI API
Input
$0.05 per voice-session minute, billed per second; backend model and tool usage billed separately
Output
$0.05 per voice-session minute, billed per second; OpenAI documents this as the voice-session price rather than separate audio input and output token rates
Low-latency live captions, realtime call transcription, microphone streams, telephony audio, and voice-interface speech recognition
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Current and generally available for realtime transcription
Input
$0.017 per minute of realtime audio
Output
Included in the per-minute realtime audio price; OpenAI does not list a separate output-token price
Local and private reasoning applications, coding assistants, agentic workflows, on-device or edge inference, fine-tuning, and cost-sensitive deployments with suitable hardware.
Context
131K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Current open-weight model; downloadable and usable through self-hosted or third-party inference infrastructure. Not served through the OpenAI API or ChatGPT.
Input
No official OpenAI API input price; self-hosting and third-party hosting costs vary.
Output
No official OpenAI API output price; self-hosting and third-party hosting costs vary.
Self-hosted reasoning, coding, agentic workflows, private deployments, research, and fine-tuning
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Current open-weight model; downloadable and deployable locally or through third-party providers; not available through the OpenAI API
Policy-based safety classification, LLM input/output filtering, content labeling, trust and safety review, and self-hosted moderation workflows
Context
131K
Reasoning
8/10
Speed
6/10
Status
Research preview; currently available as an open-weight model
Custom-policy safety classification, LLM input and output filtering, trust and safety labeling, nuanced moderation review, and offline safety analysis
Context
131K
Reasoning
8/10
Speed
3/10
Status
Research preview; open-weight and downloadable
Low-latency speech-to-speech voice agents, realtime customer support, education, accessibility, and conversational applications with function calling
Context
32K
Reasoning
5/10
Speed
9/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Status
Deprecated; scheduled for API shutdown on January 20, 2027
Input
Text: $4.00 per 1M tokens; cached text: $0.40 per 1M tokens; audio: $32.00 per 1M tokens; cached audio: $0.40 per 1M tokens; image: $5.00 per 1M tokens; cached image: $0.50 per 1M tokens
Output
Text: $16.00 per 1M tokens; audio: $64.00 per 1M tokens
Low-latency speech-to-speech voice agents, customer support, realtime assistants, and audio applications that need function calling.
Context
32K
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Status
Active and currently available
Input
$4.00 per 1M text tokens; $32.00 per 1M audio tokens; $5.00 per 1M image tokens. Cached input: $0.40 per 1M text or audio tokens and $0.50 per 1M image tokens.
Output
$16.00 per 1M text tokens; $64.00 per 1M audio tokens.
Reasoning voice agents, speech-to-speech applications, customer support, live assistants, tool-driven workflows, and long conversational sessions
Context
128K
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Streaming
Input
Text: $4.00 per 1M tokens; cached text: $0.40 per 1M; audio: $32.00 per 1M tokens; cached audio: $0.40 per 1M; image: $5.00 per 1M tokens; cached image: $0.50 per 1M
Output
Text: $24.00 per 1M tokens; audio: $64.00 per 1M tokens
Low-latency speech-to-speech agents, customer-service voice workflows, realtime tool use, telephony, and multimodal assistants with image input
Context
128K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Streaming
Input
Text: $4.00 per 1M tokens; cached text: $0.40 per 1M; audio: $32.00 per 1M audio tokens; cached audio: $0.40 per 1M; image: $5.00 per 1M tokens; cached image: $0.50 per 1M
Output
Text: $24.00 per 1M tokens; audio: $64.00 per 1M audio tokens
Low-latency spoken translation, multilingual calls, live interpretation, broadcasts, meetings, lessons, video rooms, captions, and translated audio experiences.
Context
16K
Reasoning
3/10
Speed
9/10
Capabilities
Audio input
Streaming
Input
$0.034 per minute of realtime audio
Output
$0.034 per minute of realtime audio
Low-latency live transcription, captions, meeting notes, call analysis, voice-agent input, and continuous speech-to-text workflows
Context
16K
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Current and available through the OpenAI Realtime API for realtime transcription
Input
$0.017 per minute of audio
Output
Included in the audio-duration transcription price; no separate text-output price documented
Cost-sensitive realtime voice agents, speech-to-speech applications, interactive assistants, and multimodal interfaces
Context
32K
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Status
Deprecated; currently accessible but scheduled for API shutdown on 2027-01-20
Input
$0.60 per 1M text input tokens; $0.06 per 1M cached text input tokens
Output
$2.40 per 1M text output tokens
Lower-cost, low-latency realtime voice agents, speech-to-speech assistants, and tool-enabled conversational applications
Context
128K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Multimodal input
Tool use
Input
Text: $0.60 per 1M tokens; cached text: $0.06 per 1M; audio: $10.00 per 1M tokens; cached audio: $0.30 per 1M; image: $0.80 per 1M tokens; cached image: $0.08 per 1M
Output
Text: $2.40 per 1M tokens; audio: $20.00 per 1M tokens
Governed biology, genomics, medicinal chemistry, protein analysis, drug discovery, literature synthesis, wet-lab troubleshooting, and scientific tool workflows
Reasoning
9/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Status
Generally available to eligible organizations through the trusted-access program; approved internal life sciences research only
Input
$5 per 1M input tokens; $0.50 per 1M cached input tokens
Output
$25 per 1M output tokens
High-accuracy transcription of recorded audio, streamed file transcripts, multilingual recordings, and domain-specific speech with keyword or language hints
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Streaming
Input
$0.0045 per audio minute
Output
No separate output-token price; included in the per-minute transcription price
Existing ChatGPT image-generation and image-editing integrations
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Deprecated; currently accessible; scheduled for shutdown on 2026-12-01
Input
Text: $5.00 per 1M tokens; cached text: $1.25 per 1M tokens; image: $8.00 per 1M tokens; cached image: $2.00 per 1M tokens. Per-image generation: low $0.009-$0.013, medium $0.034-$0.05, high $0.133-$0.20 depending on size.
Output
Text: $10.00 per 1M tokens; image: $32.00 per 1M tokens. Per-image generation: low $0.009-$0.013, medium $0.034-$0.05, high $0.133-$0.20 depending on size.
API-based image generation, image editing, reference-image workflows, inpainting, marketing assets, e-commerce imagery, and visual content production
Capabilities
Image input
Multimodal input
Status
Deprecated; currently accessible and scheduled to shut down on 2026-10-23
Input
Text input: $5.00 per 1M tokens; cached text input: $1.25 per 1M tokens; image input: $10.00 per 1M image tokens; cached image input: $2.50 per 1M image tokens
Output
Image generation per image: low $0.011 at 1024x1024 or $0.016 at 1024x1536 and 1536x1024; medium $0.042 or $0.063; high $0.167 or $0.25. Image output tokens: $40.00 per 1M tokens.
Production image generation, image editing, branded graphics, ecommerce product imagery, marketing assets, and workflows requiring preservation of important visual details
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Deprecated; currently accessible with API shutdown scheduled for 2026-12-01
Input
$5.00 per 1M text tokens; $8.00 per 1M image tokens; cached input $1.25 per 1M text tokens and $2.00 per 1M image tokens
Output
$10.00 per 1M text tokens; $32.00 per 1M image tokens; image generation $0.009-$0.20 per image depending on quality and resolution
High-quality text-to-image generation, reference-based image editing, text-heavy visual assets, product imagery, marketing creatives, and production design workflows.
Capabilities
Image input
Multimodal input
Status
Active; GPT Image 2.5 models are available for newer workflows, but GPT-Image-2 remains accessible as a documented API model.
Input
$8.00 per 1M image input tokens; $2.00 per 1M cached image input tokens; $5.00 per 1M text input tokens; $1.25 per 1M cached text input tokens
Output
$30.00 per 1M image output tokens; $10.00 per 1M text output tokens where applicable
Cost-sensitive image generation and editing, high-volume variations, rapid ideation, previews, lightweight personalization, and draft creative assets.
Capabilities
Image input
Multimodal input
Status
Deprecated; currently accessible but scheduled for API shutdown on 2026-12-01.
Input
Text input: $2.00 per 1M tokens; cached text input: $0.20 per 1M tokens. Image input: $2.50 per 1M image tokens; cached image input: $0.25 per 1M image tokens.
Output
Image output: $8.00 per 1M image tokens. Per-image generation: $0.005-$0.036 at 1024x1024 and $0.006-$0.052 at 1024x1536 or 1536x1024, depending on quality.
Fast, high-quality image generation and editing, creator content, product experiences, visual search, rapid prototyping, and high-volume workflows
Capabilities
Image input
Multimodal input
Status
Current; available through the OpenAI API
Input
Text input: $5.00 per 1M tokens; cached text input: $1.25 per 1M tokens; image input: $8.00 per 1M image tokens; cached image input: $2.00 per 1M image tokens
Output
Image output: $30.00 per 1M image tokens; text output is not billed because the model outputs images
Complex reasoning, advanced coding, mathematics, science, technical research, visual analysis and multi-step tool workflows
Context
200K
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current canonical alias; o3-2025-04-16 snapshot deprecated and scheduled for API shutdown on December 11, 2026
Input
$2.00 per 1M input tokens; $0.50 per 1M cached input tokens. Batch: $1.00 input and $0.25 cached input per 1M tokens.
Output
$8.00 per 1M output tokens. Batch: $4.00 per 1M output tokens.
Complex reasoning, mathematics, science, coding analysis, visual reasoning, and high-accuracy multi-step tasks
Context
200K
Reasoning
9/10
Speed
4/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Deprecated; still documented in the OpenAI API model catalog
Input
$15.00 per 1M input tokens; $7.50 per 1M cached input tokens
Output
$60.00 per 1M output tokens
Historically, difficult mathematics, science, coding, and other multi-step reasoning tasks requiring extended deliberation
Context
128K
Reasoning
9/10
Speed
4/10
Capabilities
Tool use
Streaming
Status
Retired; API access shut down on 2025-07-28
Input
$15.00 per 1 million input tokens; cached input $7.50 per 1 million tokens
Output
$60.00 per 1 million output tokens
Cost-sensitive mathematics, science, algorithmic programming, debugging, and text-only reasoning
Context
128K
Reasoning
7/10
Speed
8/10
Input
$1.10 per 1M input tokens; $0.55 per 1M cached input tokens
Output
$4.40 per 1M output tokens
Complex reasoning, difficult technical analysis, advanced programming, research workflows, and tasks where answer consistency matters more than latency or cost.
Context
200K
Reasoning
9/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Status
Deprecated in OpenAI's current model catalog; the dated snapshot o1-pro-2025-03-19 is also marked deprecated. No exact shutdown date for the canonical o1-pro alias was found in the reviewed official documentation.
Input
$150 per 1 million input tokens
Output
$600 per 1 million output tokens
Complex multi-step research, source synthesis, legal and scientific analysis, market research, and large-scale internal-data investigation
Context
200K
Reasoning
10/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$10.00 per 1M input tokens; $2.50 per 1M cached input tokens
Output
$40.00 per 1M output tokens
Coding, mathematics, science, technical analysis, structured extraction, text-to-SQL, and multi-step reasoning
Context
200K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current canonical alias with deprecated snapshot; o3-mini-2025-01-31 is scheduled for API shutdown on 2026-10-23
Input
$1.10 per 1M input tokens; $0.55 per 1M cached input tokens
Output
$4.40 per 1M output tokens
High-reliability reasoning, advanced mathematics, scientific analysis, complex coding, research, and multi-step professional work.
Context
200K
Reasoning
10/10
Speed
3/10
Capabilities
Image input
Multimodal input
Tool use
Status
Current canonical alias; the dated snapshot o3-pro-2025-06-10 is marked deprecated in the model documentation.
Input
$20 per 1 million input tokens
Output
$80 per 1 million output tokens
Fast, cost-sensitive reasoning; coding; mathematics; visual analysis; structured extraction; high-volume tool-using agents
Context
200K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Fine-tuning
Status
Deprecated; currently available through the API; scheduled for shutdown on 2026-10-23
Input
$1.10 per 1 million input tokens; $0.275 per 1 million cached input tokens
Output
$4.40 per 1 million output tokens
Complex multi-step research, source synthesis, market analysis, legal or scientific research, and long-form evidence-based reports.
Context
200K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current canonical alias; the dated snapshot o4-mini-deep-research-2025-06-26 is deprecated.
Input
$2.00 per 1M input tokens; $0.50 per 1M cached input tokens
Output
$8.00 per 1M output tokens
Text and image safety classification, content filtering, AI-output screening, policy enforcement, and human-review routing
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Text and image safety classification, content filtering, moderation queues, policy enforcement, and generated-content screening
Capabilities
Image input
Multimodal input
Status
Current default moderation model
Input
Free through the Moderation API
Output
Free through the Moderation API
Rapid video concepting, social clips, image-to-video experiments, prototypes, rough cuts, and audiovisual creative iteration
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Deprecated; currently accessible through the API as of September 23, 2026, with shutdown scheduled for September 24, 2026
Input
Not token-priced; image and text inputs are included in video-generation requests
Output
$0.10 per generated video second for 720x1280 portrait or 1280x720 landscape output
Production-quality text-to-video and image-guided video generation, cinematic prototypes, marketing assets, and high-resolution short clips with synchronized audio.
Capabilities
Image input
Multimodal input
Status
Legacy; deprecated; currently accessible through September 23, 2026; scheduled for API shutdown on September 24, 2026
Input
$0.30 per second at 720p; $0.50 per second at 1024p; $0.70 per second at 1080p. Batch pricing: $0.15, $0.25, and $0.35 per second respectively.
Output
Video with synchronized audio; pricing is charged per generated second rather than per text or audio token.
High-quality semantic search, multilingual retrieval, RAG, recommendations, clustering, classification and similarity matching
Context
8K
Reasoning
1/10
Speed
8/10
Status
Current and available through the OpenAI API
Input
$0.13 per 1 million input tokens
Output
No separate output-token charge; the model returns embedding vectors
Cost-efficient semantic search, retrieval-augmented generation, clustering, recommendations, anomaly detection, and text or code similarity
Input
$0.02 per 1 million input tokens
Output
Not applicable; embedding output is billed through input-token usage
Legacy semantic search, retrieval, clustering, recommendations, anomaly detection, and classification systems already built around ada-002 vectors
Context
8K
Reasoning
1/10
Speed
8/10
Status
Older embedding model; currently listed and accessible through the embeddings API
Input
$0.10 per 1 million input tokens
Legacy text-only safety classification and historical moderation integrations
Reasoning
1/10
Speed
8/10
Status
Retired; access ended October 27, 2025
Input
Free through the Moderation API
Output
Free through the Moderation API
Low-latency text-to-speech, realtime-oriented voice interfaces, narration, accessibility, and automated audio generation
Reasoning
1/10
Speed
9/10
Status
Current and accessible; optimized for low-latency text-to-speech
Input
$15.00 per 1 million characters
High-quality text-to-speech generation, narration, accessibility audio, voice interfaces, and downloadable speech content
Reasoning
1/10
Speed
7/10
Status
Current; available through the OpenAI Audio API speech endpoint
Input
$30 per 1 million characters
Output
Included in speech-generation pricing; output is billed by input characters rather than output tokens
Multilingual audio transcription, English speech translation, language identification, subtitles, captions, and word- or segment-level timestamps.
Capabilities
Audio input
Streaming
Status
Deprecated; currently accessible through the OpenAI API and scheduled for shutdown on February 26, 2027.
Input
$0.006 per minute of audio
Cost-sensitive cross-modal retrieval, image and video search, multimedia catalog indexing, and vector search
Context
1K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Status
Current and available through Alibaba Cloud Model Studio International deployment
Input
Image/video: $0.03 per 1 million input tokens; text: $0.09 per 1 million input tokens
Output
Free; embedding output is not charged
Qwen
Multimodal Embedding
Cross-modal retrieval, text-to-image search, image similarity, video search, semantic classification, clustering, and multimodal vector indexing.
Context
512
Reasoning
0/10
Speed
0/10
Capabilities
Image input
Video input
Multimodal input
Status
Current and accessible through Alibaba Cloud Model Studio in the China (Beijing) region; free trial pricing is listed.
Agent-environment simulation, tool-interaction modeling, terminal and software-engineering trajectories, and research on language world models
Context
262K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Current open-weight model
Low-latency duplex voice assistants, real-time customer service, AI companions, and streamed speech-to-speech applications
Context
41K
Reasoning
6/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Tool use
Web search
Streaming
Status
Current and accessible; standard-edition real-time duplex speech model
Input
Singapore: text $0.80/M tokens; audio $6.40/M tokens. China (Beijing): text $0.688/M tokens; audio $5.501/M tokens.
Output
Singapore: text $6.40/M tokens; audio $24/M tokens. China (Beijing): text $5.501/M tokens; audio $20.628/M tokens.
Low-latency voice assistants, customer service, AI companions, full-duplex spoken interaction, and voice applications using tools or cloned voices.
Context
262K
Reasoning
4/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Web search
Streaming
Status
Current and available
Input
Singapore: $0.80 per 1M text-input tokens; $6.40 per 1M audio-input tokens. China (Beijing): $0.688 per 1M text-input tokens; $5.501 per 1M audio-input tokens.
Output
Singapore: $6.40 per 1M text-output tokens; $24 per 1M audio-output tokens. China (Beijing): $5.501 per 1M text-output tokens; $20.628 per 1M audio-output tokens.
Low-latency voice assistants, real-time customer service, duplex speech conversations, interactive voice agents, and applications requiring streaming audio responses.
Context
41K
Reasoning
6/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Web search
Streaming
Status
Available; current API access confirmed, but newer models are recommended for some new projects
Input
International/Singapore: text input $0.23 per 1M tokens; audio input $0.93 per 1M tokens. China (Beijing): text input $0.413 per 1M tokens; audio input $4.126 per 1M tokens.
Output
International/Singapore: text-only output $0.70 per 1M tokens; combined text and audio output $1.87 per 1M tokens. China (Beijing): text-only output $4.126 per 1M tokens; combined text and audio output $13.752 per 1M tokens.
Real-time multilingual speech transcription, live captions, meeting transcription, voice interfaces, streaming subtitles, and Chinese-dialect recognition.
Context
8K
Reasoning
1/10
Speed
9/10
Capabilities
Audio input
Streaming
Status
Current and available through Alibaba Cloud Model Studio in the International/Singapore and China (Beijing) regions.
Input
International/Singapore: USD 0.93 per 1 million input tokens; China (Beijing): USD 0.848 per 1 million input tokens.
Output
International/Singapore: USD 0.70 per 1 million output tokens; China (Beijing): USD 0.636 per 1 million output tokens.
Expressive text-to-speech, audiobooks, film and video dubbing, content creation, premium voice services, multilingual speech, dialect synthesis, and voice cloning
Reasoning
1/10
Speed
8/10
Status
Current and available
Input
USD 0.20 per 10,000 input characters in Singapore/International; USD 0.19253 per 10,000 input characters in China (Beijing)
Output
Not separately billed; pricing is based on input characters
Long-form offline transcription of meetings, interviews, calls, media files, and multilingual or dialect-rich recordings
Context
8K
Reasoning
2/10
Speed
7/10
Status
Current and publicly available
Input
USD 0.15 per 1 million input tokens in Singapore; USD 0.113 per 1 million input tokens in China (Beijing)
Output
USD 0.47 per 1 million output tokens in Singapore; USD 0.382 per 1 million output tokens in China (Beijing)
Autonomous-driving research, driving-scene VQA, 3D BEV perception, trajectory prediction, and embodied-AI experimentation
Reasoning
6/10
Speed
5/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current open-weight research release
Text-to-image generation, image editing, text rendering in images, photorealistic scenes, creative design, and producing multiple image variants.
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current; accelerated model; functionally equivalent to qwen-image-2.0-2026-03-03
Input
$0 per input image; only generated output images are billed
Output
$0.035 per image internationally; $0.028671 per image in China (Beijing) and Singapore
Cost-sensitive text-to-image generation, posters, marketing graphics, illustrations, and images containing Chinese or English text
Status
Current and accessible; currently equivalent to qwen-image
Input
Not applicable; image generation is billed per output image
Output
$0.03/image internationally; $0.028671/image in China (Beijing)
Natural-language single-image editing, bilingual text changes, object insertion or removal, style transfer, pose changes, and image fusion
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current and accessible
Input
$0.045 per image in Singapore for international deployment
High-quality image editing, multi-image composition, industrial design concepts, geometric transformations, character-consistent edits, and controlled visual revisions
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current and available; canonical model ID is functionally equivalent to qwen-image-edit-max-2026-01-16
Input
$0.075 per output image in the International/Singapore region; $0.071677 per output image in China (Beijing)
Output
$0.075 per output image in the International/Singapore region; $0.071677 per output image in China (Beijing)
Professional text-to-image generation, image editing, posters, infographics, multilingual in-image text, photorealistic scenes, and reference-based creative production
Reasoning
6/10
Speed
5/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current rolling model; functionally equivalent to qwen-image-2.0-pro-2026-04-22
Output
$0.075 per image internationally; $0.071676 per image in China (Beijing)
General-purpose text generation, long-context analysis, multilingual applications, structured business workflows, function-calling agents, and applications that need optional reasoning mode.
Context
1M
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Web search
Streaming
Status
Current; qwen-plus currently resolves to the qwen-plus-2025-12-01 snapshot
Input
$0.40 per 1M input tokens for 0–256K input; $1.20 per 1M input tokens above 256K in the International Singapore pricing tier. International Global pricing is $0.115 per 1M tokens for up to 128K, $0.345 for 128K–256K, and $0.689 for 256K–1M.
Output
$1.20 per 1M non-thinking output tokens and $4.00 per 1M thinking output tokens for 0–256K input; $3.60 non-thinking and $12.00 thinking per 1M output tokens above 256K in the International Singapore pricing tier. International Global pricing differs by r
Complex image and video understanding, document analysis, chart interpretation, visual question answering, and structured extraction
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Currently accessible legacy visual language model; current qwen-vl-max endpoint is functionally equivalent to qwen-vl-max-2025-08-13
Input
$0.229 per 1M tokens in China (Beijing) and Singapore; $0.80 per 1M tokens for international deployment
Output
$0.573 per 1M tokens in China (Beijing) and Singapore; $3.20 per 1M tokens for international deployment
Self-hosted chat assistants, multilingual text generation, coding and mathematics assistance, long-context document work, structured text generation, and cost-sensitive private deployments
Context
131K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; publicly available for download and self-hosted deployment
Self-hosted multilingual assistants, document processing, RAG, coding support, structured extraction, and cost-conscious production deployments.
Context
131K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Legacy open-weight model; downloadable and self-hostable, while Qwen2.5 API models marked deprecated are no longer callable through current Alibaba Cloud Model Studio pricing documentation.
Input
Not applicable for the downloadable checkpoint; no current official per-token hosted price verified for this exact model.
Output
Not applicable for the downloadable checkpoint; no current official per-token hosted price verified for this exact model.
Self-hosted assistants, multilingual text generation, long-document processing, coding support, structured extraction, RAG, and agent applications
Context
131K
Reasoning
7/10
Speed
5/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available; open-weight model
Self-hosted multilingual assistants, coding, mathematics, document analysis, structured text generation, and long-context workloads
Context
131K
Reasoning
8/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight model; legacy relative to newer Qwen generations
Multimodal assistants, audio and video understanding, visual question answering, voice interaction, speech instruction following, and local multimodal AI research
Context
33K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Current; open-weight model and available through Alibaba Cloud Model Studio
Input
International: $0.10 per 1M text tokens; $6.76 per 1M audio tokens; $0.28 per 1M image/video tokens. China Beijing: $0.087 per 1M text tokens; $5.448 per 1M audio tokens; $0.287 per 1M image/video tokens.
Output
International: $0.40 per 1M tokens for text-only input; $0.84 per 1M tokens for text after multimodal input; $13.51 per 1M tokens for text-and-audio output. China Beijing: $0.345 per 1M tokens for text-only input; $0.861 per 1M tokens for text after multi
Local or self-hosted image and video understanding, OCR, document extraction, chart and diagram analysis, visual question answering, visual grounding, and multimodal research
Context
33K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Streaming
Status
Open-weight and downloadable; supported as a fine-tuning base model in Alibaba Cloud Model Studio; current hosted inference availability and standard pricing for this exact model are not clearly listed in the latest Model Studio inference-pricing catalog.
High-quality image, document, chart, screenshot, OCR, visual-grounding, and video analysis; multimodal agents and self-hosted experimentation
Context
131K
Reasoning
8/10
Speed
3/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Status
Available open-weight model; older Qwen2.5-VL generation and still listed by Alibaba Cloud Model Studio
Fast, high-volume text generation; long-context analysis; summarization; extraction; structured outputs; and applications needing optional reasoning.
Context
1M
Reasoning
7/10
Speed
9/10
Capabilities
Tool use
Web search
Streaming
Status
Current and available; canonical qwen-flash identifier is functionally equivalent to qwen-flash-2025-07-28
Input
Tiered per 1M input tokens. China Beijing: CNY 0.15 up to 128K input, CNY 0.60 above 128K to 256K, CNY 1.20 above 256K to 1M. International Singapore: CNY 0.367 up to 256K, CNY 1.835 above 256K to 1M. Regional pricing and promotional offers may vary.
Output
Tiered per 1M output tokens. China Beijing: CNY 1.50 up to 128K input, CNY 6 above 128K to 256K, CNY 12 above 256K to 1M. International Singapore: CNY 2.936 up to 256K, CNY 14.678 above 256K to 1M. Regional pricing and promotional offers may vary.
Lightweight local assistants, offline prototypes, embedded experimentation, education, simple text generation, and resource-constrained deployments
Context
33K
Reasoning
3/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model
Efficient local inference, edge applications, multilingual chat, lightweight reasoning, coding assistance, and tool-enabled agents
Context
33K
Reasoning
7/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model
Input
No official hosted API price; open-weight model
Output
No official hosted API price; open-weight model
Local and self-hosted chat, compact reasoning, coding assistance, multilingual applications, retrieval-augmented generation, and lightweight tool-using agents
Context
33K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model
Cost-efficient reasoning, coding, multilingual assistants, local deployment, structured text generation, tool-enabled agents, and fine-tuned applications
Context
131K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current and accessible; open-weight model available for self-hosting and Alibaba Cloud Model Studio API deployment
Input
$0.072 per 1 million tokens for Global deployment; $0.18 per 1 million tokens for International deployment
Output
$0.287 per 1 million tokens for non-thinking mode and $0.717 per 1 million tokens for thinking mode in Global deployment; International pricing is $0.70 non-thinking and $2.10 thinking per 1 million tokens
Local deployment, multilingual assistants, reasoning, mathematics, coding, structured text generation, research, and cost-sensitive agent workflows.
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; also available as qwen3-14b through Alibaba Cloud Model Studio, with regional capability and pricing differences.
Input
Alibaba Cloud Model Studio: China (Beijing) $0.144 per 1 million input tokens; Singapore $0.35 per 1 million input tokens. Regional prices and deployment terms may vary.
Output
Alibaba Cloud Model Studio: China (Beijing) $0.574 per 1 million output tokens in non-thinking mode and $1.434 per 1 million output tokens in thinking mode; Singapore $1.4 per 1 million output tokens in non-thinking mode and $4.2 per 1 million output toke
Self-hosted assistants, coding, mathematical and logical reasoning, multilingual applications, long-context document processing, and agentic tool-use systems
Context
131K
Reasoning
8/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Current and accessible; open-weight release with hosted Alibaba Cloud Model Studio availability
Input
US$0.108 per 1 million input tokens in several global deployments; US$0.20 per 1 million input tokens for the international Singapore deployment
Output
US$0.431 per 1 million output tokens in several global deployments; US$0.80 per 1 million output tokens for the international Singapore deployment. Thinking output is separately priced at higher rates where exposed.
Self-hosted reasoning assistants, coding agents, mathematics, multilingual applications, structured text generation, and tool-calling workflows
Context
256K
Reasoning
9/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current; open-weight model and available through Alibaba Cloud Model Studio
Input
$0.16 per 1 million tokens in international Singapore, Germany Frankfurt, and US Virginia deployments; $0.287 per 1 million tokens in China Beijing
Output
$0.64 per 1 million tokens in international Singapore, Germany Frankfurt, and US Virginia deployments; China Beijing: $1.147 per 1 million non-thinking tokens or $2.868 per 1 million thinking tokens
Complex reasoning, mathematics, software development, multilingual applications, function calling, agentic workflows, research and self-hosted open-weight deployment
Context
131K
Reasoning
9/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current and accessible through Alibaba Cloud Model Studio; original Qwen3 open-weight release, with newer 2507 instruct and thinking variants available separately
Input
$0.287 per 1M tokens for standard input in the United States, Germany and China; $0.700 per 1M tokens in Singapore. Thinking-mode input is priced the same.
Output
$1.147 per 1M tokens for standard output in the United States, Germany and China; $2.868 per 1M tokens for thinking-mode output. Singapore pricing is $2.800 standard output and $8.400 thinking-mode output per 1M tokens.
Multilingual semantic search, RAG candidate reranking, enterprise document retrieval, knowledge-base search, and improving search-result relevance.
Context
4K
Reasoning
2/10
Speed
8/10
Status
Current and available
Input
$0.10 per 1 million input tokens for the international Singapore deployment; pricing is region-dependent. Output is free.
Output
Free
Low-cost multilingual speech-to-text, language identification, offline transcription, real-time streaming ASR, and high-throughput deployments
Context
66K
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Streaming
Status
Current, open-weight, Apache 2.0 licensed
Multilingual speech transcription, language identification, long-audio processing, and self-hosted or streaming ASR applications
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Streaming
Status
Current; open-weight and downloadable
Input
No official hosted API price identified for this exact open-weight checkpoint
Output
No separate output charge for the self-hosted checkpoint
Repository-scale coding agents, code generation, code completion, debugging, refactoring, terminal workflows, and cost-sensitive self-hosted deployments.
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current; open-weight model and available through Alibaba Cloud Model Studio
Input
USD 0.144 per 1M tokens for China Beijing input up to 32K; USD 0.216 for 32K–128K; USD 0.359 for 128K–256K. International Singapore and Frankfurt pricing is USD 0.30, USD 0.50, and USD 0.80 per 1M input tokens across the same bands.
Output
USD 0.574 per 1M tokens for China Beijing output up to 32K; USD 0.861 for 32K–128K; USD 1.434 for 128K–256K. International Singapore and Frankfurt pricing is USD 1.50, USD 2.50, and USD 4.00 per 1M output tokens across the same bands.
Large-codebase analysis, code generation, refactoring, debugging, documentation and long-context coding-agent workflows
Context
1M
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Streaming
Status
Current; canonical qwen3-coder-plus currently equivalent to qwen3-coder-plus-2025-09-23
Input
US Virginia: $0.574 per 1M tokens up to 32K input; $0.861 for 32K-128K; $1.434 for 128K-256K; $2.868 for 256K-1M. Regional pricing varies.
Output
US Virginia: $2.294 per 1M tokens up to 32K input; $3.441 for 32K-128K; $5.735 for 128K-256K; $28.671 for 256K-1M. Regional pricing varies.
Streaming translation of recorded or uploaded audio and video, multilingual subtitles, translated voice tracks, and applications requiring translated text or synthesized speech.
Context
53K
Reasoning
2/10
Speed
8/10
Capabilities
Audio input
Video input
Multimodal input
Streaming
Status
Current stable model
Input
Audio input and output are billed at 12.5 tokens per second, with audio shorter than one second billed as one second. Video usage additionally consumes video tokens based on sampled frames and resolution. Monetary rates depend on the applicable Alibaba Cl
Output
Audio output is billed at 12.5 tokens per second. Text output uses the applicable Model Studio token rate when charged separately. No standalone fixed monetary rate was verified for this exact model in the model documentation.
Real-time multilingual speech interpretation, live voice translation, conference translation, streaming media, and audiovisual translation with text or synthesized speech output
Context
53K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Legacy; still available; no longer recommended for new use
Input
China (Beijing): CNY 64 per 1M input audio tokens and CNY 8 per 1M input image tokens. Singapore: CNY 73.392 per 1M input audio tokens and CNY 9.541 per 1M input image tokens.
Output
China (Beijing): CNY 64 per 1M output text tokens and CNY 240 per 1M output audio tokens. Singapore: CNY 73.392 per 1M output text tokens and CNY 278.891 per 1M output audio tokens.
Complex reasoning, coding assistance, web-grounded agents, function calling, structured extraction, and long-context text analysis
Context
262K
Reasoning
9/10
Speed
7/10
Capabilities
Tool use
Web search
Streaming
Input
International: $1.20 per 1M input tokens up to 32K; $2.40 per 1M above 32K to 128K; $3.00 per 1M above 128K to 256K. Regional prices vary.
Output
International: $6.00 per 1M output tokens up to 32K; $12.00 per 1M above 32K to 128K; $15.00 per 1M above 128K to 256K. Regional prices vary.
Local visual assistants, OCR, document and chart analysis, image question answering, lightweight video understanding, and multimodal prototyping
Context
256K
Reasoning
6/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Status
Current open-weight model
Local image and video understanding, OCR, document extraction, visual question answering, visual coding, and lightweight multimodal agents
Context
262K
Reasoning
6/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Status
Current open-weight model; available on Hugging Face and supported for supervised fine-tuning in Alibaba Cloud Model Studio
Local or hosted image and video understanding, OCR, document extraction, visual question answering, spatial reasoning, screenshot analysis, multimodal agents, and structured data extraction
Context
262K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Status
Current; open-weight model with hosted inference available through Alibaba Cloud Model Studio
Input
$0.072 per 1 million input tokens
Output
$0.287 per 1 million output tokens
Image and video understanding, OCR, document analysis, spatial reasoning, visual coding, long-context multimodal tasks, and visual-agent applications
Context
256K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Status
Current and available; open-weight release and Alibaba Cloud Model Studio API model
Input
$0.20 per 1 million tokens for the listed international deployment; regional pricing may differ
Output
$0.80 per 1 million tokens for the listed international deployment; regional pricing may differ
Document intelligence, OCR, image and video understanding, spatial reasoning, visual coding, and visual-agent applications
Context
131K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Status
Current; open-weight checkpoint and available through Alibaba Cloud Model Studio managed inference
Input
$0.16 per 1 million input tokens in Alibaba Cloud Model Studio US (Virginia) global deployment; China Beijing pricing is $0.287 per 1 million input tokens
Output
$0.64 per 1 million output tokens in Alibaba Cloud Model Studio US (Virginia) global deployment; China Beijing pricing is $1.147 per 1 million output tokens
High-quality image and video understanding, OCR, document intelligence, visual coding, spatial reasoning, long-context multimodal analysis and visual-agent applications
Context
131K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and accessible; open-weight release and Alibaba Cloud Model Studio API availability
Input
$0.287 per 1 million tokens in the US Virginia global deployment; $0.400 per 1 million tokens in Singapore
Output
$1.147 per 1 million tokens in the US Virginia global deployment; $1.600 per 1 million tokens in Singapore
Long-context multimodal analysis, document understanding, video and image interpretation, general reasoning, coding, tool-enabled assistants, and self-hosted deployment
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Accessible; no longer recommended for new projects
Input
$0.086 per 1M input tokens for requests up to 128K input tokens in US Virginia; $0.258 per 1M input tokens for 128K-256K requests
Output
$0.688 per 1M output tokens for requests up to 128K input tokens in US Virginia; $2.064 per 1M output tokens for 128K-256K requests
Efficient multimodal assistants, coding, reasoning, long-context analysis, local deployment, and tool-using agents
Context
262K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Available; open-weight Apache 2.0 model with hosted API access through Alibaba Cloud Model Studio
Input
USD 0.057 per 1 million input tokens for up to 128K input; USD 0.229 per 1 million input tokens for 128K–256K input in the Global deployment scope. International flat pricing is USD 0.25 per 1 million input tokens.
Output
USD 0.459 per 1 million output tokens for up to 128K input; USD 1.835 per 1 million output tokens for 128K–256K input in the Global deployment scope. International flat pricing is USD 2 per 1 million output tokens.
Advanced multimodal reasoning, image and video understanding, coding, long-context analysis, document and chart interpretation, function-calling agents, and web-grounded workflows.
Context
262K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current and available through Alibaba Cloud Model Studio; released globally on February 24, 2026.
Input
$0.115 per 1M tokens for input up to 128K; $0.287 per 1M tokens for input from 128K to 256K. Pricing varies by deployment scope and region.
Output
$0.917 per 1M tokens for requests up to 128K input; $2.294 per 1M tokens for requests from 128K to 256K input. Pricing varies by deployment scope and region.
Advanced multimodal reasoning, coding, video and image understanding, long-context analysis, tool-using agents, and self-hosted open-weight deployments
Context
262K
Reasoning
9/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current; open-weight model and available through Alibaba Cloud Model Studio
Input
$0.172 per 1M tokens for input up to 128K; $0.43 per 1M tokens for input above 128K and up to 256K in Beijing, Frankfurt, and Virginia. Singapore: $0.60 per 1M tokens.
Output
$1.032 per 1M tokens for input up to 128K; $2.58 per 1M tokens for input above 128K and up to 256K in Beijing, Frankfurt, and Virginia. Singapore: $3.60 per 1M tokens.
Fast long-context text, image and video understanding; structured extraction; tool-enabled agents; web-grounded applications; and high-volume multimodal workloads.
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current and available
Input
$0.029 per 1M tokens for 0–128K input; $0.115 per 1M tokens for 128K–256K input; $0.172 per 1M tokens for 256K–1M input in US Virginia/Global pricing
Output
$0.287 per 1M tokens for 0–128K input; $1.147 per 1M tokens for 128K–256K input; $1.72 per 1M tokens for 256K–1M input in US Virginia/Global pricing
Long-context reasoning, multimodal document and video analysis, coding, structured enterprise automation, function-calling agents, and web-grounded research.
Context
1M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current; canonical rolling model identifier currently equivalent to qwen3.5-plus-2026-02-15
Input
International: $0.40 per 1M input tokens for input up to 256K; $0.50 per 1M input tokens for input above 256K and up to 1M. Regional and global pricing varies.
Output
International: $2.40 per 1M output tokens for input up to 256K; $3.00 per 1M output tokens for input above 256K and up to 1M. Regional and global pricing varies.
Fast multimodal analysis, long audio understanding, audiovisual question answering, voice assistants and spoken-response applications
Context
262K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; canonical model ID functionally equivalent to qwen3.5-omni-flash-2026-03-15
Input
Singapore international: $0.40 per 1M tokens for text/image/video input and $3.00 per 1M tokens for audio input; China Beijing: $0.30 per 1M tokens for text/image/video input and $2.48 per 1M tokens for audio input
Output
Singapore international: $2.20 per 1M tokens for text output and $11.90 per 1M tokens for text-and-audio output; China Beijing: $1.83 per 1M tokens for text output and $9.90 per 1M tokens for text-and-audio output
Low-latency voice assistants, speech-to-speech applications, realtime multimedia analysis, interactive agents, and multimodal conversations.
Context
262K
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current and available; rolling model identity functionally equivalent to qwen3.5-omni-flash-realtime-2026-03-15
Input
International: $0.55 per 1M tokens for text/image/video input; $4.50 per 1M tokens for audio input. China mainland: $0.45 per 1M tokens for text/image/video input; $3.71 per 1M tokens for audio input.
Output
International: $3.30 per 1M tokens for text output; $17.70 per 1M tokens for audio output. China mainland: $2.75 per 1M tokens for text output; $14.71 per 1M tokens for audio output.
Multilingual voice assistants, speech-enabled multimodal applications, audio-visual analysis, spoken explanations, accessibility tools, and interactive media workflows.
Context
262K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Input
$1.40 per 1M tokens for text/image/video input; $11.00 per 1M tokens for audio input in international deployments. China (Beijing) snapshot pricing is $0.96 per 1M tokens for text/image/video input and $7.29 per 1M tokens for audio input.
Output
$8.30 per 1M tokens for text output; $44.00 per 1M tokens for text-and-audio output in international deployments. China (Beijing) snapshot pricing is $5.50 per 1M tokens for text output and $29.29 per 1M tokens for text-and-audio output.
Real-time voice assistants, speech-to-speech applications, multimodal customer service, visual conversational agents, live multimedia analysis, and interactive applications requiring controllable speech output.
Context
262K
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current and available; canonical model identifier functionally equivalent to snapshot qwen3.5-omni-plus-realtime-2026-03-15
Input
China (Beijing): $1.38 per 1M text/image/video tokens; $11 per 1M audio tokens. Singapore: $2.10 per 1M text/image/video tokens; $16.50 per 1M audio tokens.
Output
China (Beijing): $8.25 per 1M text-output tokens; $41.26 per 1M text-and-audio-output tokens. Singapore: $12.40 per 1M text-output tokens; $62 per 1M text-and-audio-output tokens.
Coding agents, repository-level software engineering, visual document analysis, video understanding, STEM reasoning, long-context assistants, and self-hosted multimodal applications
Context
262K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current and available; open-weight release and Alibaba Cloud Model Studio API model
Input
$0.60 per 1M tokens internationally; $0.412564 per 1M tokens in China Beijing and Singapore
Output
$3.60 per 1M tokens internationally; $2.475384 per 1M tokens in China Beijing and Singapore
Agentic coding, long-context software engineering, multimodal analysis, tool-using agents, and self-hosted deployments
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current; open-weight and available through Alibaba Cloud Model Studio
Input
USD 0.248 per 1M tokens in US Virginia, Germany Frankfurt, and China Beijing; USD 0.375 per 1M tokens in Singapore
Output
USD 1.485 per 1M tokens in US Virginia, Germany Frankfurt, and China Beijing; USD 2.25 per 1M tokens in Singapore
Fast multimodal assistants, coding agents, visual document analysis, video understanding, tool-using workflows, object localization, and large-context applications
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Input
$0.25 per 1M input tokens up to 256K input tokens; $1.00 per 1M input tokens above 256K and up to 1M, Singapore international pricing
Output
$1.50 per 1M output tokens up to 256K input tokens; $4.00 per 1M output tokens above 256K and up to 1M, Singapore international pricing
Advanced coding agents, front-end development, long-context analysis, structured API workflows, and text generation with web search
Context
262K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Web search
Streaming
Status
Preview; currently accessible; scheduled for deprecation on October 10, 2026
Input
$1.30 per 1M tokens for input up to 128K; $2.00 per 1M tokens for input above 128K and up to 256K, international pricing
Output
$7.80 per 1M tokens for requests up to 128K input; $12.00 per 1M tokens for requests above 128K and up to 256K input, international pricing
Long-context multimodal analysis, agentic coding, OCR, object localization, frontend development, visual reasoning, and tool-enabled enterprise assistants
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current; rolling model ID qwen3.6-plus is currently equivalent to qwen3.6-plus-2026-04-02
Input
$0.276 per 1M tokens for 0-256K input tokens; $1.101 per 1M tokens for 256K-1M input tokens in Global deployment
Output
$1.651 per 1M tokens for 0-256K input tokens; $6.602 per 1M tokens for 256K-1M input tokens in Global deployment
Fast multimodal agents, visual coding, tool-use workflows, search agents, and long-context document or screen analysis
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current and available; rolling qwen3.7-flash identifier is currently equivalent to qwen3.7-flash-2026-07-15
Input
Global: $0.028 per 1M input tokens for 0-32K; $0.083 for 32K-256K; $0.165 for 256K-1M. Singapore international: $0.030, $0.100, and $0.200 per 1M input tokens respectively.
Output
Global: $0.110 per 1M output tokens for 0-32K; $0.330 for 32K-256K; $0.660 for 256K-1M. Singapore international: $0.130, $0.400, and $0.800 per 1M output tokens respectively.
Long-context reasoning, multimodal document and video analysis, coding, tool-using agents, structured extraction, and enterprise productivity workflows
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Current; rolling identifier currently equivalent to qwen3.7-plus-2026-05-26
Input
$0.40 per 1M tokens up to 256K input tokens and $1.20 per 1M tokens from 256K to 1M input tokens in US Virginia; regional and promotional pricing varies
Output
$1.60 per 1M tokens up to 256K input tokens and $4.80 per 1M tokens from 256K to 1M input tokens in US Virginia; thinking-mode output uses the same listed rates
Multilingual semantic search, retrieval-augmented generation, code retrieval, recommendation, clustering, classification, and large-scale text vectorization
Status
Current and available through Alibaba Cloud Model Studio internationally
Input
$0.07 per 1 million input tokens in Singapore international deployment; China pricing may differ by region and billing mode
Advanced reasoning, coding, scientific and professional research, long-context analysis, and long-horizon agent workflows
Context
1M
Reasoning
10/10
Speed
4/10
Capabilities
Tool use
Web search
Streaming
Status
Current and available; open-weight release and hosted API model
Input
$2 per 1 million tokens internationally; $1.65 per 1 million tokens in China (Beijing)
Output
$6 per 1 million tokens internationally; $4.951 per 1 million tokens in China (Beijing)
Coding assistants, repository analysis, visual document workflows, long-context research, office automation, multimodal agents, and tool-using applications
Context
1M
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Active and currently available; open-weight release and Alibaba Cloud Model Studio API model
Input
$0.424 per 1M tokens in China Beijing; $0.50 per 1M tokens in Singapore. Implicit cache input is $0.085 per 1M tokens in Beijing and $0.10 per 1M tokens in Singapore. Explicit cache creation is $0.53 per 1M tokens in Beijing and $0.625 per 1M tokens in Si
Output
$1.696 per 1M tokens in China Beijing; $3 per 1M tokens in Singapore.
Fast long-context reasoning, coding assistance, visual document and chart analysis, video understanding, function-calling agents, and high-concurrency applications
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and accessible through Alibaba Cloud Model Studio
Input
$0.113 per 1 million input tokens; implicit cache input $0.014 per 1 million tokens; explicit cache creation $0.177 per 1 million tokens; explicit cache read $0.014 per 1 million tokens
Output
$0.382 per 1 million output tokens
High-volume coding, agentic workflows, long-context document and codebase analysis, office automation, and multimodal text-image-video understanding
Context
262K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Fine-tuning
Status
Current open-weight experimental preview
Complex coding, autonomous software engineering, long-horizon agent workflows, professional document analysis, visual reasoning, long videos, and demanding research tasks.
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Web search
Status
Stable official release; currently available through Alibaba Cloud Model Studio
Input
US$1.65 per 1M input tokens and US$2.00 per 1M input tokens for International scope; regional pricing varies. Cached-input pricing starts at US$0.206 per 1M tokens for the listed regional/global scope and US$0.25 per 1M tokens for International scope.
Output
US$4.951 per 1M output tokens for the listed regional/global scope and US$6.00 per 1M output tokens for International scope.
Real-time speech translation, multilingual meetings, live interpretation, translated voice communication, and audiovisual translation with low latency.
Context
53K
Reasoning
4/10
Speed
9/10
Capabilities
Image input
Audio input
Multimodal input
Streaming
Status
Current stable model
Input
Singapore: audio input $7.50 per 1 million tokens; image input $0.55 per 1 million tokens. China (Beijing): audio input $5.653 per 1 million tokens; image input $0.466 per 1 million tokens.
Output
Singapore: text output $20 per 1 million tokens; audio output $30 per 1 million tokens. China (Beijing): text output $14.133 per 1 million tokens; audio output $22.613 per 1 million tokens.
Long-form audio and video understanding, multimedia analysis, audio-visual agents, content summarization, and tool-using workflows
Context
1M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current and available
Input
USD 0.15 per 1 million input tokens for International deployment; USD 0.016 per 1 million cache-hit input tokens
Output
USD 0.47 per 1 million output tokens for International deployment
Real-time voice assistants, speech-to-speech applications, interactive video agents, live media analysis, multimodal customer service, meeting and collaboration interfaces, and applications requiring tool or MCP integration.
Context
197K
Reasoning
7/10
Speed
9/10
Capabilities
Audio input
Video input
Multimodal input
Tool use
Streaming
Status
Current and available; international deployment in Singapore
Input
China (Beijing): CNY 1.5 per 1 million tokens for text/images/video input and CNY 6 per 1 million tokens for audio input. Singapore: CNY 1.677 per 1 million tokens for text/images/video input and CNY 6.781 per 1 million tokens for audio input.
Output
China (Beijing): CNY 4.5 per 1 million tokens for text output and CNY 12 per 1 million tokens for audio output. Singapore: CNY 5.104 per 1 million tokens for text output and CNY 13.636 per 1 million tokens for audio output.
Real-time moderation of streamed prompts and language-model responses
Context
8K
Reasoning
3/10
Speed
8/10
Status
Available open-weight model
Low-latency character dialogue, virtual companions, game NPCs, role-playing applications, IP character replication, and conversational smart devices
Context
33K
Reasoning
4/10
Speed
9/10
Capabilities
Web search
Streaming
Status
Current; dynamically updated managed model
Input
Regional pricing: USD 0.034 per 1M input tokens in Beijing and US Virginia; USD 0.05 per 1M input tokens in Singapore. Cached input is listed at USD 0.007 per 1M tokens in Beijing and US Virginia and USD 0.01 in Singapore. Alibaba Cloud also publishes loc
Output
Regional pricing: USD 0.203 per 1M output tokens in Beijing and US Virginia; USD 0.40 per 1M output tokens in Singapore. Alibaba Cloud also publishes localized regional prices.
Complex text-to-image layouts, multilingual typography, posters, menus, storyboards, interface mockups, product visuals, and image editing with one to three reference images
Context
5K
Reasoning
2/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current and available
Input
$0.003 per image in the international Singapore deployment
Output
$0.04 per 1K image or $0.075 per 2K image in the international Singapore deployment; regional pricing varies
Instruction-based image editing, multi-image fusion, character-consistent compositions, object replacement, style transfer, poster and text editing, and product-image variation.
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current; canonical model currently equivalent to qwen-image-edit-plus-2025-10-30
Input
0; image editing is billed by output image
Output
$0.028671 per image in China (Beijing); $0.03 per image in Singapore/international
Realistic text-to-image generation, creative concepts, marketing visuals, editorial artwork, product concepts, and general-purpose image creation.
Reasoning
1/10
Speed
5/10
Status
Current and available for text-to-image generation; the undated qwen-image-max identifier is currently equivalent to qwen-image-max-2025-12-30.
Input
No separate input charge; text prompt input is included in per-image billing.
Output
$0.075/image in Singapore international deployment; $0.071677/image in China (Beijing).
Fast text-to-image generation, text-heavy layouts, image editing, reference-image compositing, marketing graphics, and high-volume image workflows.
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through Alibaba Cloud Model Studio
Input
$0.003 per input image internationally; regional pricing varies
Output
$0.03 per generated image internationally; regional pricing varies
Complex multimodal analysis, long-context document understanding, image and video question answering, audio-aware workflows, coding, and enterprise applications requiring broad input support.
Context
128K
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Available/listed for API pricing, but not identified as a baseline model that is always publicly available; account-dependent availability should be verified.
Input
$2.00 per 1M input tokens; $0.02 per image; $0.08 per video minute; $0.02 per audio minute
Output
$6.00 per 1M output tokens
Low-latency image and video understanding, object detection, physical AI, robotics, edge deployment, and on-device visual assistants
Context
16K
Reasoning
7/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Streaming
Status
Current and publicly available
Input
$0.10 per 1 million tokens; $0.005 per image; $0.03 per video minute
Output
$0.10 per 1 million tokens
Fast multimodal applications, document and image analysis, short-video understanding, structured extraction, multilingual chat, coding, and tool-using agents
Context
128K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current and publicly available baseline model
Input
$0.80 per 1 million input tokens; $0.01 per image; $0.06 per minute of video; $0.015 per minute of audio
Output
$2.00 per 1 million output tokens
Low-latency reasoning, coding assistance, function calling, local deployment, on-device applications and cost-sensitive inference
Context
66K
Reasoning
8/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Current and accessible; open-weight research-preview model with API availability through Reka Gateway
Input
$0.10 per 1 million tokens through Reka Gateway
Output
$0.20 per 1 million tokens through Reka Gateway
Coding, mathematical reasoning, local inference, research experimentation, and fine-tuning for agentic workflows
Context
33K
Reasoning
7/10
Speed
7/10
Status
Available as an open-weight model; official API availability was announced but the current public baseline API catalog does not list the exact model identifier
Audio-driven digital humans, lip-sync video, singing avatars, multilingual character animation, multi-person dialogue, and long-duration talking-video generation
Reasoning
1/10
Speed
9/10
Capabilities
Image input
Audio input
Multimodal input
Status
Current; ongoing project
Visual question answering, high-resolution image understanding, multimodal search, image-grounded research, and tool-assisted agentic reasoning
Context
262K
Reasoning
8/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Status
Current open-weight research model
Visual deep-search, fine-grained image understanding, multimodal agent research, and self-hosted tool-using vision-language applications
Context
262K
Reasoning
8/10
Speed
4/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Status
Current open-weight model
Spatial intelligence research, image-text question answering, visual reasoning, 3D scene analysis, and solid-geometry problems
Context
33K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current open-weight release
Native image generation, high-resolution visual creation, image editing, infographic and layout generation, visual understanding, and multimodal research or creative workflows
Reasoning
6/10
Speed
4/10
Capabilities
Image input
Multimodal input
Status
Current open-weight flagship checkpoint
Input
No official hosted API price verified; downloadable model weights are available
Output
No official hosted API price verified; downloadable model weights are available
Unified computer-vision research, detection, OCR, segmentation, depth and normal estimation, visual grounding, and multi-view geometry
Reasoning
7/10
Speed
4/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight research model
Long-horizon multimodal agents, data analysis, deep research, complex information presentation, office automation, and tool-driven workflows
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Preview; currently available through the SenseNova Token Plan
Input
Free during the public beta Token Plan; standard per-token API pricing is not publicly verified
Output
Free during the public beta Token Plan; standard per-token API pricing is not publicly verified
Character-consistent image generation for multi-episode videos, motion comics, short dramas, storyboards, and cross-shot visual production
Capabilities
Image input
Multimodal input
Status
Current; integrated into the SenseNova Seko series and Seko 2.0 platform
Professional image creation, infographics, advertising, e-commerce assets, presentations, educational diagrams, storyboards, and multi-step visual delivery workflows
Reasoning
8/10
Speed
5/10
Capabilities
Image input
Multimodal input
Status
Current production release; enterprise API available through whitelist access
Open-source visual understanding, image generation, image editing, infographic creation, visual reasoning, and continuous image-text workflows.
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Open-source model series; U1-8B-MoT and U1-A3B-MoT variants are available. SenseNova U1.5 is the newer successor for visual creation and editing, while U1 checkpoints remain documented and downloadable.
Fast infographic generation, dense visual explanations, charts, diagrams, presentation graphics, and information-heavy layouts
Status
Current and accessible through SenseNova Token Plan and dedicated image-generation integrations
Input
Free public beta with usage quota; commercial pricing not verified
Output
Free public beta with usage quota; commercial pricing not verified
Text-to-image generation, reference-image creation, visual design, posters, infographics, product imagery, and iterative image editing
Reasoning
5/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through the SenseNova Token Plan
Coding agents, software engineering, long-context reasoning, tool-using agents, private local inference, and work-centric automation
Context
262K
Reasoning
9/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Current; open-weight model available for local deployment and accessible through StepFun's API platform
High-throughput coding agents, visual document and UI understanding, search-heavy research workflows, long-context analysis, and tool-using autonomous agents
Context
256K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Status
Current and available
Input
$0.20 per 1 million tokens for cache misses; $0.04 per 1 million tokens for cache hits
Output
$1.15 per 1 million tokens
Long-running agentic workflows, software engineering, coding, professional knowledge work, financial analysis, research, and vision-assisted tasks
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Status
Current preview; available through StepFun products and API. Open-weight release planned for October 15, 2026.
Input
¥7 per 1 million uncached input tokens; ¥0.35 per 1 million cached input tokens
Output
Â¥20 per 1 million output tokens
Multilingual transcription, meetings, subtitles, live content, customer-service recordings, specialized terminology, noisy audio, dialects, code-switching, and singing or music transcription
Reasoning
5/10
Speed
9/10
Capabilities
Audio input
Streaming
Status
Current and available through the StepFun API
Input
2.8 CNY per audio hour
Zero-shot text-to-speech, natural-language voice design, singing and vocal generation, music, sound effects, ambience, and complete multi-element audio scenes
Capabilities
Audio input
Multimodal input
Status
Current research and platform-listed model; public API and access details are limited
Song generation, instrumental music, lyric-to-song workflows, accompaniment, cover-style synthesis, and rapid music prototyping
Reasoning
2/10
Speed
3/10
Capabilities
Audio input
Multimodal input
Status
Current; publicly showcased and available through StepFun's StepAudio 3 music experience
Natural realtime voice conversation, full-duplex interaction, interruption-aware assistants, emotional audio understanding and voice agents that use tools.
Reasoning
7/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Streaming
Status
Current and publicly listed
Controllable multilingual text-to-speech, narration, voice interfaces, localization, and expressive spoken-audio generation
Reasoning
1/10
Speed
8/10
Status
Current and publicly accessible
On-device visual assistants, GUI grounding, screen understanding, spatial reasoning, automotive interaction, and low-latency local agent workflows
Context
131K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Status
Current; officially announced as an edge-deployment base model
On-device speech recognition, audio understanding, voice assistants, in-vehicle interaction, privacy-sensitive audio processing, and low-latency edge AI.
Reasoning
6/10
Speed
8/10
Status
Current; edge-deployment model with public capability information but no verified public hosted API pricing or general model-download documentation located.
On-device text-to-image generation, local image editing, privacy-sensitive creative features, and low-latency edge-device applications.
Reasoning
1/10
Speed
9/10
Capabilities
Image input
Multimodal input
Status
Current research model; public technical information is available, but official public API availability and downloadable weights were not verified.
Low-latency desktop and mobile GUI automation, visual grounding, local computer-use agents, and privacy-sensitive edge workflows
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Status
Current; edge-deployment GUI agent model
Self-hosted text generation, research, quantization, and domain-specific fine-tuning
Context
2K
Reasoning
4/10
Speed
7/10
Status
Available open-weight model; older Falcon generation with newer successors
Input
No official hosted API price; self-hosted model weights
Output
No official hosted API price; self-hosted model weights
Research, self-hosted text generation, model fine-tuning, summarization, and general language-model experimentation
Context
2K
Reasoning
4/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available as open weights; legacy-generation model
Input
No official hosted API pricing; self-hosted open weights
Output
No official hosted API pricing; self-hosted open weights
Research, self-hosted text generation, language-model evaluation, and domain-specific fine-tuning when substantial GPU infrastructure is available
Context
2K
Reasoning
6/10
Speed
3/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model; legacy-generation checkpoint
Input
No official first-party hosted API pricing; self-hosted/open-weight distribution
Output
No official first-party hosted API pricing; self-hosted/open-weight distribution
Memory-efficient local text generation, edge-device experimentation, research, and downstream fine-tuning
Context
33K
Reasoning
3/10
Speed
8/10
Status
Current open-weight model; downloadable from Hugging Face
Input
No official hosted API pricing; self-hosted model weights
Output
No official hosted API pricing; self-hosted model weights
Efficient local, edge, and resource-constrained English text generation; experimentation with BitNet models; full fine-tuning on the supplied prequantized revision.
Context
33K
Reasoning
3/10
Speed
8/10
Status
Current open-weight/downloadable model
Memory-efficient local text generation, edge deployment, BitNet research, continued pretraining, and custom fine-tuning
Context
33K
Reasoning
4/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; publicly downloadable
Lightweight local text generation, continued pretraining, domain adaptation, language-model research, and edge-oriented deployments
Context
16K
Reasoning
3/10
Speed
8/10
Status
Current open-weight model
Lightweight local chat, text generation, edge deployment, prototyping, and fine-tuning experiments
Context
16K
Reasoning
2/10
Speed
9/10
Capabilities
Fine-tuning
Streaming
Status
Current; open-weight; instruction-tuned
Local multilingual text generation, long-context experimentation, research, and downstream fine-tuning
Context
131K
Reasoning
6/10
Speed
8/10
Status
Current open-weight model; available for download and self-hosted inference
Efficient local text generation, multilingual language modeling, long-context applications, model adaptation, and compact reasoning research.
Context
131K
Reasoning
6/10
Speed
8/10
Status
Current open-weight model
Input
No official hosted API pricing; downloadable weights are available for self-hosted deployment.
Output
No official hosted API pricing; downloadable weights are available for self-hosted deployment.
Efficient local inference, compact conversational assistants, multilingual text generation, long-context applications, and resource-constrained deployments
Context
131K
Reasoning
7/10
Speed
8/10
Status
Current; open-weight
Input
No official hosted API pricing; self-hosted/open-weight model
Output
No official hosted API pricing; self-hosted/open-weight model
Efficient local or private deployment, multilingual instruction following, compact conversational applications, coding assistance, mathematics, and long-context text processing
Context
131K
Reasoning
5/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; downloadable and usable for self-hosted inference
Local text generation, multilingual research, domain adaptation, fine-tuning, long-context experiments, and resource-conscious deployment.
Context
131K
Reasoning
4/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; downloadable from Hugging Face and usable with supported local inference frameworks.
Input
No official hosted API pricing; self-hosted model weights
Output
No official hosted API pricing; self-hosted model weights
Local multilingual chat, lightweight instruction following, retrieval-augmented generation, efficient text generation, and cost-sensitive self-hosted deployments
Context
131K
Reasoning
5/10
Speed
8/10
Status
Current open-weight model
Private or local multilingual text generation, long-context experimentation, domain adaptation, and fine-tuning from a 7B-scale open-weight foundation
Context
262K
Reasoning
6/10
Speed
8/10
Status
Current open-weight model; pretrained base checkpoint
Input
No official hosted API pricing; downloadable open-weight checkpoint
Output
No official hosted API pricing; deployment cost depends on self-hosted infrastructure
Self-hosted multilingual assistants, long-context text processing, coding, research, private deployments, and quantized local inference
Context
262K
Reasoning
7/10
Speed
8/10
Status
Current open-weight model
Self-hosted multilingual text generation, long-context research, coding, RAG, foundation-model experimentation, and downstream fine-tuning
Context
262K
Reasoning
7/10
Speed
5/10
Status
Current open-weight model; available from the official Hugging Face repository
Long-context text generation, multilingual instruction following, document processing, retrieval-augmented generation, coding and self-hosted enterprise deployments
Context
262K
Reasoning
7/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model; publicly downloadable and usable with Transformers, vLLM and llama.cpp
Input
No official hosted API price; self-hosted model weights
Output
No official hosted API price; self-hosted model weights
Arabic NLP, dialect-aware assistants, long-context document analysis, summarization, multilingual reasoning, and locally deployed applications
Reasoning
7/10
Speed
7/10
Status
Current; open model family
Ultra-lightweight local text generation, edge deployment, offline instruction following, rewriting, extraction, and embedded AI experiments
Context
262K
Reasoning
2/10
Speed
10/10
Capabilities
Fine-tuning
Streaming
Status
Current open-weight model
Input
No official hosted API price; downloadable weights
Output
No official hosted API price; downloadable weights
Lightweight local Python generation, fill-in-the-middle completion, edge deployment, code education, and low-resource developer tools
Context
262K
Reasoning
2/10
Speed
9/10
Status
Current and publicly accessible open-weight model
Lightweight local text generation, instruction-following, edge devices, embedded applications, experimentation, and privacy-sensitive deployments.
Context
262K
Reasoning
2/10
Speed
9/10
Status
Current open-weight model
Input
No official hosted API pricing found; model weights are available for local deployment under the Falcon-LLM License.
Output
No official hosted API pricing found; local inference costs depend on hardware and deployment framework.
Lightweight local reasoning, edge deployment, offline assistants, experimentation, and compact text-generation applications
Context
262K
Reasoning
7/10
Speed
8/10
Status
Current open-weight model; downloadable and available for self-hosted inference
Lightweight local function calling, edge automation, API argument generation, and resource-constrained assistants
Context
262K
Reasoning
2/10
Speed
9/10
Capabilities
Tool use
Streaming
Status
Current downloadable open-weight model
Local reasoning experiments, edge deployment, offline assistants, privacy-sensitive text processing, and resource-constrained inference
Context
262K
Reasoning
4/10
Speed
9/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight checkpoint
Ultra-lightweight local reasoning, embedded applications, edge deployment, experimentation, and low-memory text generation
Context
262K
Reasoning
3/10
Speed
9/10
Status
Current open-weight model
Mathematical reasoning, programming, long-context analysis, local deployment, self-hosted inference, and test-time scaling
Context
262K
Reasoning
8/10
Speed
8/10
Status
Current open-weight model
Research, multilingual text generation, fine-tuning, quantization, and self-hosted inference
Context
8K
Reasoning
4/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight pretrained base model
Input
No official hosted API price; downloadable weights for self-hosted deployment
Output
No official hosted API price; infrastructure-dependent
Local inference, multilingual text completion, research, continued pretraining, domain adaptation, and fine-tuning on constrained hardware
Context
4K
Reasoning
4/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight pretrained base model
Input
No official hosted API price; self-hosted/open-weight model
Output
No official hosted API price; self-hosted/open-weight model
Lightweight local assistants, multilingual instruction following, extraction, classification, education, and resource-conscious deployments
Context
8K
Reasoning
4/10
Speed
8/10
Capabilities
Tool use
Streaming
Status
Current open-weight model; downloadable from Hugging Face
Input
No official first-party hosted API pricing
Output
No official first-party hosted API pricing
Fine-tuning, multilingual text generation, research, compact local inference, and edge-oriented deployments
Context
8K
Reasoning
5/10
Speed
8/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API price; downloadable weights for self-hosted deployment
Output
No official hosted API price; downloadable weights for self-hosted deployment
Self-hosted chat assistants, multilingual text generation, lightweight coding and mathematics, local inference, and resource-conscious deployments
Context
32K
Reasoning
6/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available as an open-weight model
Fine-tuning, multilingual text generation, code and mathematics experiments, research, and self-hosted inference
Context
33K
Reasoning
6/10
Speed
7/10
Status
Current open-weight model; downloadable and self-hostable
Self-hosted multilingual chat, instruction following, reasoning, mathematics, coding, research, and long-context text generation
Context
33K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API price; downloadable weights
Output
No official hosted API price; downloadable weights
Fine-tuning, multilingual text generation, language-model research, code and mathematics experimentation, and self-hosted inference
Context
33K
Reasoning
7/10
Speed
6/10
Capabilities
Fine-tuning
Streaming
Status
Available open-weight model
Input
No official hosted API pricing; self-hosted/open-weight deployment
Output
No official hosted API pricing; self-hosted/open-weight deployment
Self-hosted multilingual assistants, STEM and mathematics tasks, coding, instruction following, research, and local function-calling systems.
Context
33K
Reasoning
7/10
Speed
5/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Available open-weight model; official repository remains accessible. No official deprecation or shutdown date found.
Input
No official hosted API pricing; downloadable weights are provided for self-managed inference.
Output
No official hosted API pricing; downloadable weights are provided for self-managed inference.
Local English text generation, open-weight research, long-sequence experimentation, and memory-conscious inference
Context
8K
Reasoning
4/10
Speed
8/10
Status
Available for download; legacy or superseded by Falcon-H1-7B-Base
Local or self-hosted document OCR, formula recognition, table extraction, receipts, invoices, papers, and layout-aware document parsing
Context
16K
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Streaming
Status
Current open-weight release
Natural-language object grounding, open-vocabulary detection, promptable instance segmentation, crowded-scene perception, robotics and visual inspection pipelines
Context
8K
Reasoning
2/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current; open-weight; Apache-2.0 licensed
Open-source text-to-speech, reference-voice generation, speech and lyric editing, emotion and timbre transformation, speech enhancement, and source separation
Reasoning
2/10
Speed
5/10
Capabilities
Audio input
Multimodal input
Fine-tuning
Status
Current open-source release
Lightweight local assistants, edge-oriented inference, long-context text processing, prototyping, quantized deployment, and low-resource instruction-following workloads.
Context
262K
Reasoning
3/10
Speed
9/10
Status
Current open-weight release
Efficient local text generation, long-context analysis, lightweight reasoning, coding assistance, and agent-oriented applications
Context
256K
Reasoning
6/10
Speed
9/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; publicly available for download and local deployment
Local and self-hosted text generation, Chinese and multilingual instruction following, long-context analysis, mathematics, coding, reasoning, and lightweight agent workloads
Context
262K
Reasoning
7/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current open-weight model; publicly available for self-hosted deployment
Input
No official hosted API price published; self-hosted model weights are available
Output
No official hosted API price published; self-hosted inference costs depend on infrastructure
Chinese and multilingual text generation, reasoning, coding, long-context workloads, private local inference, quantized deployment, and domain fine-tuning.
Context
262K
Reasoning
7/10
Speed
7/10
Status
Current open-weight downloadable model
Input
No first-party hosted API token price identified; downloadable weights are intended for self-hosted or third-party deployment.
Output
No first-party hosted API token price identified; downloadable weights are intended for self-hosted or third-party deployment.
Open-weight reasoning, long-context analysis, mathematics, science, coding, agent workflows, and cost-conscious self-hosted inference
Context
262K
Reasoning
8/10
Speed
8/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current and accessible as an open-weight model; also listed as the Tencent Cloud API model hunyuan-a13b. Tencent Cloud documentation notes an ongoing migration of Hunyuan services toward TokenHub.
Input
Â¥0.50 per 1 million input tokens on Tencent Cloud postpaid API
Output
Â¥2 per 1 million output tokens on Tencent Cloud postpaid API
Large-scale Chinese and English text generation, reasoning, mathematics, coding, long-context analysis, research, and self-hosted experimentation
Context
256K
Reasoning
8/10
Speed
4/10
Capabilities
Fine-tuning
Streaming
Status
Available as an open-weight release; legacy for the historical Tencent Cloud hunyuan-large API identifier
Self-hosted multilingual translation, localization, translation research, and Chinese dialect or minority-language translation
Context
33K
Reasoning
2/10
Speed
5/10
Status
Current open-weight model
Image-to-3D asset creation, game and virtual-world content, product visualization, design prototyping, and self-hosted 3D generation
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current open-weight release; self-hosted deployment
Input
No official hosted token or per-request API price; model weights are available for self-hosting
Output
No official hosted output price; generated 3D assets are produced through self-hosted inference
Local image-to-3D asset generation, textured mesh creation, game and design prototypes, Blender workflows, and research on open 3D generative models
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available open-weight model system; superseded by the newer Hunyuan3D-2.1 release but still publicly accessible
Text-to-3D, image-to-3D, sketch-to-3D, rapid game and e-commerce asset creation, 3D printing, and production-oriented asset prototyping.
Capabilities
Image input
Multimodal input
Status
Current and accessible through Tencent Cloud APIs; newer HY-3D-3.1 is also available as a separate model version.
Input
Credit-based pricing: 25 credits for a default Professional normal textured generation; 15 credits for Geometry mode; 30 credits for LowPoly mode. Additional features such as PBR, multi-view, and custom face count consume extra credits.
Output
Not token-priced; generated 3D assets are billed by generation credits.
Local English- and Chinese-language text-to-image generation, creative prototyping, ComfyUI workflows, LoRA customization, and ControlNet-based image conditioning
Reasoning
1/10
Speed
6/10
Status
Open-weight and downloadable; no official deprecation or shutdown notice found
Input
No official hosted API input pricing published; intended primarily for local deployment
Output
No official hosted API output pricing published; image generation uses local compute rather than a documented token price
Text-to-image generation, reference-guided image creation, marketing graphics, e-commerce content, visual ideation, and image-generation applications requiring custom aspect ratios
Capabilities
Image input
Multimodal input
Status
Current and accessible through Tencent Cloud TokenHub; also available as an open-weight HunyuanImage-3.0 release
Output
Approximately 0.20 CNY per generated image in Tencent Cloud mainland China TokenHub pricing; approximately 0.032 USD per image in cited international pricing
Multilingual OCR, document parsing, text spotting, table and formula extraction, structured information extraction, and local visual-document processing
Context
131K
Reasoning
3/10
Speed
8/10
Capabilities
Image input
Multimodal input
Fine-tuning
Streaming
Status
Current open-weight release
Video description, video question answering, video summarization, content review, scene analysis, and video metadata generation.
Context
32K
Reasoning
3/10
Speed
8/10
Capabilities
Video input
Multimodal input
Status
Online and currently listed in Tencent Cloud TokenHub; the older Hunyuan platform entry was retired on 2026-06-22, while the TokenHub model remains available.
Input
CNY 3 per 1 million input tokens
Output
CNY 9 per 1 million output tokens
Research and production experimentation with high-quality local text-to-video generation
Reasoning
1/10
Speed
3/10
Status
Open-source and currently accessible; original HunyuanVideo model, with HunyuanVideo-1.5 released later as a lighter successor
Local text-to-video and image-to-video generation, open-source video research, creative prototyping, and developers needing a comparatively lightweight high-quality video model.
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current and publicly accessible open-weight model
Input
No official hosted API price identified; model weights are available for local deployment under the Tencent Hunyuan Community License.
Output
No official hosted API price identified; local inference costs depend on hardware and infrastructure.
Image-to-video generation, reference-image animation, visual effects, creative video prototyping, and locally hosted open-weight video workflows
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Available as an open-weight model and official open-source repository; no official shutdown date found
Image-grounded reasoning, OCR, chart and document analysis, visual localization, educational problem solving, and multilingual visual question answering
Context
40K
Reasoning
8/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Online and currently available through Tencent Cloud TokenHub
Input
CNY 3 per 1M input tokens
Output
CNY 9 per 1M output tokens
Coding agents, long-context analysis, complex reasoning, productivity automation, structured workflows, and multi-step tool use
Context
256K
Reasoning
8/10
Speed
7/10
Capabilities
Tool use
Web search
Fine-tuning
Status
Current; open-weight and available through Tencent Cloud TokenHub
Input
CNY 1 per 1 million input tokens; cached input CNY 0.25 per 1 million tokens on Tencent Cloud TokenHub
Output
CNY 4 per 1 million output tokens on Tencent Cloud TokenHub
Automated decomposition of FBX 3D models into separate model components.
Status
Current and API-accessible
Input
Not publicly specified in the model documentation; enabling post-processing adds 20 points.
Fast text-to-3D and image-to-3D asset generation, prototyping, and automated 3D content pipelines
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
15–25 Tencent Cloud points per generation request; 1 point is listed as CNY 0.12
Automated conversion of existing 3D assets between common interchange and delivery formats
Status
Current TokenHub 3D format-conversion service
Generating short human character animations from natural-language action descriptions
Reasoning
1/10
Speed
6/10
Capabilities
Multimodal input
Status
Current and accessible through Tencent TokenHub
Input
10 points per generation request
Automated retopology, polygon reduction, and preparation of existing 3D meshes for games, rendering, animation, and downstream asset workflows
Status
Current and API-accessible
Automated rigging and skinning of human or animal 3D characters for animation, games, virtual characters, and asset prototyping
Reasoning
1/10
Speed
5/10
Status
Current and accessible through Tencent Cloud TokenHub
Input
10 Tencent Cloud credits per request
Output
Included in the 10-credit per-request charge; output is a rigged 3D model file
Reference-guided texturing of existing OBJ or GLB meshes, including PBR material generation and automated asset preparation
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub API
Output
30 Tencent Cloud points per generation
Automated UV unwrapping and preparation of 3D assets for texturing, rendering, game development, and digital-content workflows.
Reasoning
1/10
Speed
7/10
Status
Current and accessible through Tencent Cloud TokenHub
Output
10 points per call; Tencent Cloud states that 1 point equals CNY 0.12
Real-time and asynchronous Mandarin, English, mixed-language, and Chinese-dialect transcription, captions, subtitles, and short voice-command recognition
Reasoning
4/10
Speed
8/10
Capabilities
Audio input
Streaming
Status
Preview; internal-test availability
Input
Usage-based Tencent Cloud ASR pricing; exact model price not verified in the reviewed official documentation
Output
No separate output price; transcription is billed through the applicable Tencent Cloud ASR or large-model 2.0 pricing scheme
High-resolution text-to-image generation, reference-image creation, multi-turn image editing, posters, UI concepts, marketing assets, and product-image workflows
Capabilities
Image input
Multimodal input
Web search
Status
Current preview model
Input
10 CNY per million tokens; reference-image generation uses the published TokenHub token rules
Output
15,000 tokens per 1K or 2K image; 20,000 tokens per 4K image, equivalent to approximately CNY 0.15 and CNY 0.20 respectively at the listed CNY 10 per million-token rate
Low-latency multilingual translation, localized content, structured translation instructions, and cost-sensitive production workflows
Context
8K
Reasoning
2/10
Speed
9/10
Status
Current and available through Tencent Cloud TokenHub
Input
Â¥0.3 per 1 million tokens
Output
Â¥1.2 per 1 million tokens
Professional multilingual translation, localization, terminology-sensitive workflows, and context-aware business translation
Context
8K
Reasoning
2/10
Speed
8/10
Status
Current and available through Tencent Cloud TokenHub
Input
Â¥0.5 per 1 million tokens in China; US$0.074 per 1 million tokens on Tencent Cloud international pricing
Output
Â¥2 per 1 million tokens in China; US$0.295 per 1 million tokens on Tencent Cloud international pricing
Professional, domain-specific and high-quality multilingual translation with contextual disambiguation and instruction following
Context
8K
Reasoning
6/10
Speed
8/10
Status
Online and currently available through Tencent Cloud TokenHub
Input
0.5 CNY per 1 million input tokens
Output
2 CNY per 1 million output tokens
Chinese role-play, character simulation, fictional dialogue, AI avatars, and emotionally oriented conversational experiences
Context
32K
Reasoning
3/10
Speed
7/10
Status
Current and available through Tencent Cloud TokenHub
Input
CNY 2.4 per 1 million input tokens
Output
CNY 9.6 per 1 million output tokens
Image understanding, OCR, chart and diagram analysis, STEM visual reasoning, visual question answering, and multi-image comparison
Context
44K
Reasoning
7/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
Â¥7.5 per 1 million input tokens
Output
Â¥17.5 per 1 million output tokens
Generating explorable 3D environments, Gaussian-splat scenes, point clouds, and collision meshes from text or reference images
Reasoning
1/10
Speed
2/10
Capabilities
Image input
Multimodal input
Status
Current and accessible through Tencent Cloud TokenHub
Input
Â¥10 per 1 million tokens; reference usage is 8,000,000 tokens per generated scene
Output
Approximately ¥80 per generated 3D scene at the documented reference usage
Text-to-3D, image-to-3D, multi-view reconstruction, game assets, product visualization, digital-human content, e-commerce assets and 3D printing workflows
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud HY-3D APIs and TokenHub
Input
Not token-priced; text and image inputs are included in the per-generation task charge
Output
15–60 credits per generation; CNY 0.12 per credit, approximately CNY 1.80–7.20 per generation
Long-context coding agents, complex tool-use workflows, productivity automation, document analysis, game development, and scientific reasoning
Context
1M
Reasoning
8/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Preview; currently available as an open-weight model and through Tencent products, Tencent Cloud TokenHub, and OpenRouter
Input
CNY 6 per 1M tokens; cached input CNY 0.3 per 1M tokens
Output
CNY 18 per 1M tokens
Text-to-panorama and image-to-panorama generation for immersive environments, virtual tours, games, simulations, visualization, and 3D-world pipelines.
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Current and accessible through Tencent Cloud TokenHub
Input
USD 1.60 per million tokens; reference usage 481,250 tokens per panorama
Output
Approximately USD 0.77 per generated panorama
High-volume semantic retrieval, vector search, FAQ matching, text clustering, classification, and cost- or latency-sensitive knowledge-base applications
Context
33K
Reasoning
1/10
Speed
8/10
Status
Current and available through Tencent Cloud TokenHub
Input
$0.07 per million text-input tokens
High-quality multilingual semantic search, retrieval-augmented generation, enterprise knowledge bases, similarity matching, and text classification.
Context
32K
Reasoning
1/10
Speed
8/10
Status
Current and available through Tencent Cloud TokenHub and the Tencent Cloud embeddings API.
Input
USD 0.084 per million input tokens internationally; RMB 0.6 per million input tokens on the China TokenHub pricing page.
Output
Not applicable; the model is billed for text input tokens and returns embedding vectors.
Fast cross-modal image-text retrieval, multimodal semantic matching, and video search
Context
33K
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Video input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
USD 0.07 per million text-input tokens; USD 0.098 per million image-input tokens; USD 0.21 per million video-input tokens
High-precision multimodal retrieval, cross-modal image-text search, video search, and semantic matching across text, image, and video collections
Context
33K
Reasoning
1/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
USD 0.084 per million text-input tokens; USD 0.126 per million image-input tokens; USD 0.252 per million video-input tokens
Video, image, audio, and text understanding; multimedia summarization; content tagging; structural video analysis; object localization; enterprise media workflows
Context
128K
Reasoning
5/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Streaming
Status
Currently accessible; scheduled for shutdown on October 15, 2026 at 00:00 Beijing time on Tencent Cloud TokenHub and ADP
Input
Â¥1.2 per million tokens
Output
Â¥3.5 per million tokens
Multilingual video translation, voice-preserving dubbing, subtitle translation, online courses, films, and short-form video localization.
Reasoning
2/10
Speed
5/10
Capabilities
Audio input
Video input
Multimodal input
Input
CNY 10 per 1 million tokens; usage is resolution-dependent and billed according to video duration and token consumption.
Multilingual video localization, translated online courses, short-form video dubbing, film and media localization, and long-video voice-preserving translation
Reasoning
2/10
Speed
6/10
Capabilities
Video input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub as an asynchronous AI dubbing model
Input
¥0.2061 per second for 720p or lower; ¥0.2311 per second at 1080p; ¥0.2811 per second at 2K or higher. TokenHub lists the underlying rate as ¥10 per million tokens.
Fast everyday image creation, social-media graphics, marketing materials, short-video covers, and reference-guided image generation
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
10 CNY per million tokens; input images are free for the Flash tier
Output
Reference consumption: 45,000 tokens per 1K image (0.45 CNY), 67,500 tokens per 2K image (0.675 CNY), and 100,800 tokens per 4K image (1.008 CNY)
Low-cost, high-volume text-to-image and reference-to-image generation, including e-commerce product imagery, batch visual assets, and marketing content
Reasoning
1/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Currently available managed image-generation model
Input
10 CNY per 1 million tokens; reference pricing is approximately 0.162 CNY per 1K image, 0.18 CNY per 2K image, and 0.225 CNY per 4K image
Output
Approximately 0.162 CNY per 1K image, 0.18 CNY per 2K image, and 0.225 CNY per 4K image
Professional text-to-image and reference-to-image generation, brand visuals, refined product imagery, high-quality design assets, and 1K-to-4K creative production.
Capabilities
Image input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
Reference images: first 3 images free; from the 4th image, 0.10 CNY per image
Output
1K: 0.95 CNY per image; 2K: 0.95 CNY per image; 4K: 1.71 CNY per image
Fast, cost-conscious generation of short e-commerce, advertising, social media, and reference-guided video assets
Capabilities
Image input
Audio input
Video input
Multimodal input
Status
Current and accessible through Tencent Cloud TokenHub
Audio and video transcription, subtitle generation, sentence-level timestamps, and short-form speech-to-text workflows
Capabilities
Audio input
Video input
Multimodal input
Status
Current and available through Tencent Cloud TokenHub
Input
10 CNY per 1 million tokens; reference usage is 50 tokens per second, approximately 0.0005 CNY per second
Multi-view and video-to-3D reconstruction, camera and depth estimation, point-map prediction, surface-normal estimation, novel-view synthesis, and 3D Gaussian Splatting workflows
Reasoning
2/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Legacy but publicly available open-weight model; superseded by WorldMirror-2.0 in Tencent's HY-World 2.0 catalog
Long-context analysis, enterprise agents, research, coding assistance, structured extraction and tool-enabled workflows
Context
1M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$1.25 per 1M tokens below 200K prompt tokens; $2.50 per 1M tokens at or above 200K prompt tokens
Output
$2.50 per 1M tokens below 200K prompt tokens; $5.00 per 1M tokens at or above 200K prompt tokens
Software engineering, codebase analysis, technical reasoning, long-context work, tool-using agents, document analysis, and structured workflow automation
Context
500K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$2.00 per 1M tokens; $0.30 per 1M cached tokens; long context of 200K tokens or more: $4.00 input and $0.60 cached input per 1M tokens
Output
$6.00 per 1M tokens; long context of 200K tokens or more: $12.00 per 1M tokens
Advanced software engineering, long-context reasoning, agentic tool use, research with web or X search, and professional knowledge work.
Context
500K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current; available through the xAI API, Grok Build, Cursor, and selected model gateways.
Input
$2.00 per 1 million input tokens below 200,000 prompt tokens; $0.50 per 1 million cached input tokens; $4.00 per 1 million input tokens above 200,000 prompt tokens; $1.00 per 1 million cached input tokens above 200,000 prompt tokens.
Output
$6.00 per 1 million output tokens below 200,000 prompt tokens; $12.00 per 1 million output tokens above 200,000 prompt tokens.
Agentic coding, long-context software engineering, research, knowledge work, visual analysis, structured extraction, and tool-using workflows.
Context
500K
Reasoning
9/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Current and available; superseded as xAI's flagship by Grok 4.7 but still supported on the xAI API.
Input
$2.00 per 1M tokens below 200k prompt tokens; $4.00 per 1M tokens at 200k tokens or more. Cached input is $0.50 per 1M tokens below 200k and $1.00 per 1M tokens at 200k or more.
Output
$6.00 per 1M tokens below 200k prompt tokens; $12.00 per 1M tokens at 200k tokens or more.
Low-latency coding, interactive development, and agentic workflows in Cursor or Grok Build
Context
500K
Reasoning
9/10
Speed
10/10
Capabilities
Image input
Multimodal input
Tool use
Status
Available as a faster-serving variant in Cursor and Grok Build; not available as a separate public xAI API model
Input
2x standard Grok 4.7 input-token rate where Fast pricing applies; standard Grok 4.7 API rate is $2 per 1M input tokens
Output
2x standard Grok 4.7 output-token rate where Fast pricing applies; standard Grok 4.7 API rate is $6 per 1M output tokens
Deep research, parallel investigation, long-context analysis, and tool-assisted synthesis
Context
1M
Reasoning
9/10
Speed
5/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Status
Beta; currently available through the xAI API
Input
$1.25 per 1M tokens; $2.50 per 1M tokens for requests at or above 200K input tokens
Output
$2.50 per 1M tokens; $5.00 per 1M tokens for requests at or above 200K input tokens
Fast general-purpose text generation, image-aware analysis, coding assistance, structured extraction, tool-calling agents and large-context workflows
Context
1M
Reasoning
6/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$1.25 per 1M tokens; $0.20 per 1M cached tokens. Long-context requests at or above 200K prompt tokens: $2.50 per 1M input tokens; $0.40 per 1M cached tokens.
Output
$2.50 per 1M tokens; $5.00 per 1M tokens for long-context requests at or above 200K prompt tokens.
Complex reasoning, coding, technical research, long-context document analysis, image understanding, structured responses, and tool-enabled agentic workflows
Context
1M
Reasoning
8/10
Speed
8/10
Capabilities
Image input
Multimodal input
Tool use
Web search
Streaming
Input
$1.25 per 1M tokens for prompts under 200K tokens; $2.50 per 1M tokens for prompts at or above 200K tokens; cached input is $0.20 or $0.40 per 1M tokens respectively
Output
$2.50 per 1M tokens for prompts under 200K tokens; $5.00 per 1M tokens for prompts at or above 200K tokens
Agentic coding, web development, debugging, software engineering workflows, MCP integrations, and fast tool-calling applications.
Context
256K
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Current; public beta
Input
$1.00 per 1M input tokens; $0.20 per 1M cached input tokens. Long-context pricing for prompts exceeding 200K tokens is $2.00 per 1M input tokens and $0.40 per 1M cached input tokens.
Output
$2.00 per 1M output tokens; $4.00 per 1M output tokens when the prompt exceeds 200K tokens.
API-based text-to-image generation, image editing, visual prototyping, creative applications, and per-image billing
Context
1K
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Current and available through the xAI API as of September 24, 2026
Input
$0.002 per input image; text prompt pricing is not separately stated
Output
$0.02 per generated image for 1K and 2K resolution
High-quality text-to-image generation, image editing, multi-reference compositing, marketing graphics, product imagery, typography, and design assets
Capabilities
Image input
Multimodal input
Status
Generally available
Input
$0.01 per input image
Output
$0.04-$0.08 per generated image, depending on resolution and quality
High-quality image generation and editing, realistic product and marketing imagery, detailed scenes, creative assets, and images requiring stronger text rendering or prompt adherence.
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Status
Currently available; scheduled for API retirement on November 2, 2026. After retirement, requests using the slug are served by Grok Imagine Image 2.0 with quality set to low.
Input
$0.01 per input image
Output
$0.05 per 1K image; $0.07 per 2K image
Short-form text-to-video, image-to-video, reference-guided video, cinematic prototyping, marketing clips, and audiovisual creative workflows
Capabilities
Image input
Audio input
Multimodal input
Status
Generally available
Input
$0.01 per image; preset audio input is free
Output
$0.08/sec at 480p; $0.14/sec at 720p; $0.25/sec at 1080p
Low-cost text-to-video and image-to-video drafts, social clips, rapid creative iteration, and high-volume generation
Reasoning
0/10
Speed
8/10
Capabilities
Image input
Multimodal input
Status
Current and available through the xAI Imagine API
Input
$0.01 per image input; video input is charged per second where applicable
Output
$0.02 per second at 480p; $0.03 per second at 720p; $0.14 per second at 1080p
Realtime voice agents, customer support, telephony, sales, multilingual conversations, and tool-enabled spoken workflows
Reasoning
8/10
Speed
9/10
Capabilities
Audio input
Multimodal input
Tool use
Web search
Streaming
Status
Current and available through the xAI Speech to Speech API
Input
$0.08 per minute of audio; $0.004 per text input
Output
$0.08 per minute of audio
Batch and real-time multilingual speech transcription, dictation, voice assistants, accessibility, meetings, and customer-support audio
Capabilities
Audio input
Streaming
Status
Currently accessible; original speech-to-text model; scheduled for deprecation in the coming weeks
Input
$0.10 per hour of audio for REST batch transcription; $0.20 per hour of audio for streaming
Output
Included in the transcription service price; output is text transcript data
Batch and real-time transcription of multilingual, noisy, conversational, telephony, and voice-agent audio
Reasoning
0/10
Speed
9/10
Capabilities
Audio input
Streaming
Input
$0.10 per hour of audio for batch REST transcription; $0.20 per hour of audio for streaming transcription
Few-shot audio-language research, speech continuation, voice and style conversion, speech translation, speech editing, and audio-text experimentation
Context
8K
Reasoning
5/10
Speed
3/10
Capabilities
Audio input
Multimodal input
Status
Current open-weight downloadable model
Local audio understanding, speech-to-text dialogue, spoken conversational agents, and controllable text-to-speech research
Context
8K
Reasoning
7/10
Speed
4/10
Capabilities
Audio input
Multimodal input
Status
Current open-weight model; locally downloadable and usable
Input
No official hosted API price published; self-hosted weights
Output
No official hosted API price published; self-hosted weights
Embodied AI research, autonomous-driving perception and planning, spatial reasoning, affordance prediction, robot navigation, and multimodal visual analysis
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Status
Active open-weight model
Research, custom post-training, foundation-model experimentation, and self-hosted long-context inference
Context
256K
Reasoning
8/10
Speed
9/10
Status
Current open-weight base model
Speech-to-text transcription for Chinese, English, regional Chinese dialects, code-switched speech, lyrics, noisy recordings, and multi-speaker conversations
Reasoning
3/10
Speed
7/10
Capabilities
Audio input
Streaming
Status
Current and accessible through the Xiaomi MiMo API; open-source code and weights available
Input
$0.074 per hour of input audio for overseas API usage; ¥0.5 per hour on the domestic pricing page
Output
No separate output charge documented; billing is based on input-audio duration
Expressive text-to-speech, audiobooks, podcasts, dubbing, character dialogue, voice interfaces, narrated content, and stylized speech or singing.
Context
8K
Reasoning
1/10
Speed
7/10
Input
Free for a limited time
Output
Free for a limited time
Zero-shot voice cloning, expressive narration, character dialogue, personalized speech, and custom-voice audio production
Capabilities
Audio input
Multimodal input
Streaming
Status
Current; limited-time free access
Input
Free for a limited time
Output
Free for a limited time
Custom synthetic voices for narration, characters, podcasts, ASMR, games, assistants, and creative audio production
Status
Current; limited-time free access
Input
Free for a limited time
Output
Free for a limited time
Local coding agents, terminal automation, cybersecurity experimentation, visual coding and agentic reinforcement-learning research
Context
262K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Multimodal input
Tool use
Streaming
Status
Current open-weight SFT checkpoint
Input
No official hosted API price; open-weight download
Output
No official hosted API price; open-weight download
High-volume multimodal API workloads, coding assistants, agent automation, long-context document and repository analysis, tool-using workflows, and cost-sensitive professional applications.
Context
1M
Reasoning
8/10
Speed
9/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; open-weight and available through the Xiaomi MiMo API, MiMo Studio, MiMo Desktop, MiMo Code, and other supported integrations.
Input
$0.14 per million tokens for cache-miss input; $0.0028 per million tokens for cache-hit input.
Output
$0.28 per million tokens.
Long-horizon agents, coding, cybersecurity, research, computer use, multimodal analysis, and complex multi-step workflows
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; open-source weights and API access available
Input
$0.435 per 1M tokens cache miss; $0.0036 per 1M tokens cache hit; batch input $0.2175 per 1M tokens
Output
$0.87 per 1M tokens; batch output $0.435 per 1M tokens
Latency-sensitive multimodal reasoning, coding, tool-use, long-context research, and interactive agent workflows
Context
1M
Reasoning
9/10
Speed
10/10
Capabilities
Image input
Audio input
Video input
Multimodal input
Tool use
Status
Current; high-speed inference variant of MiMo-V2.6-Pro
Input
$0.036 per million cached-input tokens; $4.35 per million uncached-input tokens
Output
$8.70 per million output tokens
Local image and video understanding, multimodal reasoning, visual question answering, OCR-oriented tasks, visual grounding, and GUI analysis
Context
128K
Reasoning
7/10
Speed
7/10
Capabilities
Image input
Video input
Multimodal input
Status
Current open-weight model; publicly available for download and local deployment
Input
No official first-party hosted API pricing found; open-weight download
Output
No official first-party hosted API pricing found; open-weight download
Self-hosted image and video understanding, multimodal reasoning research, supervised fine-tuning, and reinforcement-learning experimentation
Context
128K
Reasoning
7/10
Speed
6/10
Capabilities
Image input
Video input
Multimodal input
Fine-tuning
Status
Current open-weight checkpoint; available for download
Input
No official hosted API pricing; downloadable weights are available under the MIT license
Output
No official hosted API pricing; downloadable weights are available under the MIT license
Long-horizon agent workflows, repository-scale coding, complex software engineering, tool-driven automation, and very large documents
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Deprecated; API model name scheduled to expire on 2026-10-21 at 10:00 Beijing Time
Input
¥0.025 per million tokens cache hit; ¥3 per million tokens cache miss; equivalent USD prices $0.0036 and $0.435 per million tokens
Output
Â¥6 per million tokens; equivalent USD price $0.87 per million tokens
Russian-language conversational assistants, long-context dialogue, RAG, document analysis, information extraction, reporting, and complex business text generation
Context
131K
Reasoning
8/10
Speed
7/10
Status
Current and available in Yandex AI Studio
Input
$0.00204918 per 1,000 input tokens in asynchronous mode, before VAT
Output
$0.0083606544 per 1,000 output tokens in asynchronous mode, before VAT
High-volume business text processing, customer-support dialogs, moderation, classification, summarization, document extraction, and knowledge-base search.
Context
66K
Reasoning
5/10
Speed
9/10
Status
Current and available in Yandex AI Studio
Text-to-image generation, Russian-language visual content, illustrations, graphic design, advertising assets, presentations, landing pages, and product cards
Context
500
Reasoning
1/10
Speed
7/10
Capabilities
Image input
Multimodal input
Streaming
Input
$0.0182786856 per image-generation request, without VAT
Russian-language enterprise text generation, document analysis, RAG, structured extraction, rewriting, classification, reporting, and tool-augmented business assistants
Context
33K
Reasoning
7/10
Speed
7/10
Capabilities
Tool use
Web search
Fine-tuning
Streaming
Status
Current and available through Yandex Cloud AI Studio; explicit model URI recommended. The RC alias remains available until the end of support.
Input
0.8 RUB per 1,000 input tokens including VAT; 0.8 RUB per 1,000 cached input tokens; 0.2 RUB per 1,000 tool tokens. Asynchronous mode: 0.41 RUB per 1,000 input tokens including VAT.
Output
0.8 RUB per 1,000 output tokens including VAT. Asynchronous mode: 0.41 RUB per 1,000 output tokens including VAT.
Fast, cost-efficient Russian-language text generation and everyday conversational tasks
Reasoning
6/10
Speed
9/10
Short-form text-to-video, image animation, start-and-end-frame transitions, advertising, marketing, realistic scenes, and 3D-style video generation
Capabilities
Image input
Multimodal input
Status
Current and available through the Z.ai video-generation API
Input
$0.20 per video
Output
$0.20 per video
Long-horizon software engineering, repository-scale coding, code agents, complex refactoring, large-document analysis, research implementation, and tool-using workflows.
Context
1M
Reasoning
9/10
Speed
7/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Active; still listed in Z.AI API pricing and available as downloadable open weights, although GLM-5.3 is the newer flagship successor.
Input
$1.40 per 1 million input tokens; cached input $0.26 per 1 million tokens
Output
$4.40 per 1 million output tokens
Complex software engineering, large codebases, long-horizon coding agents, terminal workflows, technical research, and authorized cybersecurity analysis
Context
1M
Reasoning
9/10
Speed
6/10
Capabilities
Tool use
Fine-tuning
Streaming
Status
Current and available
Multimodal coding agents, screenshot and interface understanding, long-context document work, visual verification, tool-assisted research, and cost-sensitive agent workflows
Context
1M
Reasoning
9/10
Speed
9/10
Capabilities
Image input
Video input
Multimodal input
Tool use
Streaming
Status
Current and available through the Z.ai API, GLM Coding Plan, and publicly released open weights
Input
$0.15 per 1 million input tokens; cached input $0.03 per 1 million tokens
Output
$0.50 per 1 million output tokens
Hosted multilingual speech-to-text, short real-time transcriptions, captions, meeting snippets, voice input, customer-service audio, and terminology-aware transcription
Capabilities
Audio input
Streaming
Status
Current and publicly available through the Z.AI audio transcription API
Input
Approximately CNY 0.06 per minute; verify current account pricing
Output
Not separately priced; transcription is billed by audio duration according to reported listings
Text-heavy posters, presentation graphics, educational diagrams, social-media layouts, image editing, and open-weight image-generation research
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current; available through the Z.ai API and as downloadable open weights
Input
$0.015 per image
Output
$0.015 per generated image
High-volume OCR, PDF parsing, table and formula recognition, handwriting, receipts, forms, information extraction, and local document-processing pipelines
Reasoning
2/10
Speed
9/10
Capabilities
Image input
Multimodal input
Fine-tuning
Status
Current; open-weight; available through hosted API and self-hosted deployment
Input
$0.03 per 1 million tokens
Output
$0.03 per 1 million tokens
