What is Grok Imagine Image Quality?
Grok Imagine Image Quality is xAI’s higher-quality image generation and editing model. It was introduced as Quality Mode for the Grok Imagine API and is intended for projects where the visual result matters more than minimizing generation cost. A prompt can describe a new image, while an uploaded image can provide source material for an edit or transformation.
The model is not a general-purpose text model. Its output is an image rather than a written answer, and the supplied specifications do not indicate support for text generation, structured text responses, audio, video, embeddings, or general tool use. Its role is narrower and more practical: create or modify images with an emphasis on detail and fidelity.
Where it fits in the Grok Imagine lineup
Grok Imagine Image Quality belongs to xAI’s Grok Imagine family. Within that family, it represents the quality-focused option rather than a lowest-cost or highest-throughput setting. xAI positions it around more realistic output, finer textures, stronger prompt following, improved text rendering, and greater consistency for creative work.
The model has documented aliases including grok-imagine-image-quality-20260403, grok-imagine-image-quality-latest, and grok-imagine-image-pro. These aliases refer to the same model family entry and should not be treated as separate models when comparing capabilities or pricing.
Image quality and creative control
The main reason to choose this model is its emphasis on image quality. According to xAI’s positioning, Quality Mode is intended to improve realism, textures, character and scene detail, text rendering, prompt adherence, and consistency across creative outputs. These claims are provider descriptions rather than independent benchmark results, so they should be treated as positioning information rather than a guaranteed score or universal advantage.
In practical terms, the model is a better fit when a generated image needs to look finished enough for review, publication, or further editing. Examples include a product image with a specific composition, a cinematic scene with detailed materials, a social graphic containing readable wording, or an edit that needs to preserve important aspects of a source image while changing its appearance.
Higher quality does not mean that every generation will be correct. Image models can still misinterpret detailed instructions, produce unwanted visual elements, or render text imperfectly. For important commercial work, outputs should be reviewed and regenerated as necessary.
Supported inputs and outputs
The model supports both text and image inputs. Text prompts can specify the subject, setting, style, composition, or desired changes. Image inputs allow an application to use an existing image as the basis for editing or transformation.
Its output is generated or edited imagery. The documented output sizes are approximately 1K, or 1024 by 1024 pixels, and 2K, or 2048 by 2048 pixels. The supplied research does not specify a broader set of aspect ratios, a maximum number of input images, a maximum prompt length, or a maximum output count, so those details should not be assumed from the resolution information alone.
| Capability | Supported detail |
|---|---|
| Text input | Yes |
| Image input | Yes |
| Image output | Yes |
| Text output | No documented text output |
| Audio or video input | No |
| Audio or video output | No |
| Output resolutions | Approximately 1K and 2K |
| Batch API | Not supported |
API pricing
xAI’s supplied pricing information lists separate charges for input images and generated output images. Image inputs cost $0.01 per image. Image outputs cost $0.05 per 1K image and $0.07 per 2K image.
| Usage | Price |
|---|---|
| Input image | $0.01 per image |
| 1K output image | $0.05 per image |
| 2K output image | $0.07 per image |
For example, a request that uses one source image and produces one 2K result has a listed image charge of $0.08 before any other account-level considerations. A text-only generation request does not incur the input-image charge because it has no image input. The exact total for a workflow depends on how many images are submitted and generated.
The model does not support the Batch API according to the supplied documentation. It also has a documented base-tier rate limit of six requests per second. Applications with larger workloads should account for that limit and the absence of batch processing when estimating throughput.
Technical capabilities and limitations
Grok Imagine Image Quality is multimodal in the specific sense that it can accept text and images, but its output is image-only. It is not documented as a reasoning or coding model, and those capabilities are not meaningful substitutes for its image-generation function. The supplied model record gives it comparative editorial scores of 1 for reasoning and 1 for coding; these are database evaluations, not provider-published benchmarks and should not be interpreted as measurements of image quality.
The model record lists no context-length limit and no maximum-output-token limit. That is expected for an image-output model, but it does not mean that prompts or image inputs are unlimited. Because xAI’s supplied documentation does not provide those limits here, applications should consult the current API documentation before designing around unusually long prompts or complex input workflows.
There is no documented support for function calling, external tools, streaming, fine-tuning, caching, JSON mode, or batch processing. These omissions matter if an application needs a general agent, structured text responses, or high-volume asynchronous generation. Grok Imagine Image Quality should instead be treated as a focused image endpoint that an application can place inside a larger workflow.
Best use cases
- Product visualization: Create realistic product scenes, promotional compositions, or concept imagery from detailed descriptions.
- Marketing and advertising: Produce campaign concepts and graphics where visual polish and prompt adherence are more important than the lowest per-image price.
- Cinematic and realistic scenes: Generate detailed environments, characters, materials, and lighting for creative development.
- Image editing: Use a source image as input for transformations or controlled creative changes.
- Social media graphics: Create images that require more legible text and a specified visual style, while still checking the final lettering manually.
- Creative production workflows: Generate multiple high-detail candidates for human selection and refinement.
The model is especially suitable when a 2K output is useful and the additional $0.02 over a 1K output is justified by the intended use. For drafts, exploratory concepts, or very large numbers of low-stakes variations, a lower-cost or faster image option may be more appropriate if available in the current xAI catalog.
When to choose Grok Imagine Image Quality
Choose this model when image fidelity is the main decision criterion and the documented per-image pricing is acceptable. It is a strong fit for a team that needs detailed generation or editing, wants both text and image inputs, and can review outputs rather than relying on automatic perfection.
Consider another type of option when your priority is speed, cost, or scale. A lower-quality image mode may be preferable for thumbnails, early brainstorming, automated variations, or other cases where a fast inexpensive draft is enough. A model with batch support is better for large queued workloads. A general-purpose language model is more appropriate when the primary task is writing, coding, reasoning, or structured data generation rather than producing images.
Applications should also consider model longevity. The current slug is scheduled for retirement on November 2, 2026. xAI says that requests using grok-imagine-image-quality will then be served by grok-imagine-image-2.0 with quality set to low. The request and response shapes are expected to remain compatible, but relying on that redirect can change the resulting image quality. New integrations should plan an explicit migration to Grok Imagine Image 2.0 and select the desired quality behavior rather than treating the redirect as a permanent compatibility guarantee.
Availability and retirement status
Grok Imagine Image Quality is currently available through the xAI API. xAI’s retirement notice began on September 2, 2026, and the scheduled retirement date is November 2, 2026. The model’s current status and exact availability should be checked against xAI’s live documentation when deploying a new application, because the model is not intended to remain a stable long-term identifier after that date.
For an existing integration, the practical steps are to identify every use of the quality-model slug or its aliases, test the replacement model explicitly, compare 1K and 2K outputs, and verify that any application assumptions about image quality still hold. The current model remains useful for supported workflows before retirement, but its planned redirect is an important limitation for production systems that require reproducible behavior over time.