Yi Large FC

What is Yi Large FC?

Yi Large FC is a specialized language model from 01.AI, identified in the API by the model ID yi-large-fc. It belongs to the Yi Large model family but is distinguished by its focus on function calling. Rather than only returning a conversational answer, it can evaluate a request against a set of tools supplied by an application, choose an appropriate function, and return the arguments required to invoke that function.

In practical terms, a function is a defined operation in an application. Examples might include looking up an order, retrieving a weather report, querying an internal database, or creating a calendar event. The model does not independently execute those operations. Instead, it produces a structured tool call that the surrounding application can validate and execute.

This makes Yi Large FC different from a general-purpose chat model whose main output is natural-language text. Its value is in coordinating language understanding with software actions while keeping the action request in a machine-readable format.

Where it fits in 01.AI’s lineup

Yi Large FC is part of 01.AI’s Yi model family and is positioned as a developer-facing model rather than a consumer assistant. The broader 01.AI ecosystem includes foundation models, model deployment, fine-tuning, private deployment, and enterprise AI services. Within that context, Yi Large FC addresses a narrower problem: reliable tool selection and structured invocation for applications and agent workflows.

Its “FC” designation refers to function calling, not to a separate media or multimodal capability. The supplied model information classifies it as text-in and text-out. There is no verified support for image, audio, or video input, and it does not produce images, audio, or video. It should therefore be evaluated as a text model with tool-use output, not as a general multimodal model.

How its function-calling design works

An application supplies the model with descriptions of available tools, including the function names, their purposes, and the arguments they accept. Given a user request, Yi Large FC can determine whether one of those functions is relevant. If so, it returns a structured call containing the selected function and its arguments.

For example, an application might expose a check_order_status function that accepts an order number. A user could ask where an order is, and the model could identify the function and provide the order number in the expected argument structure. The application would then execute the function and may send the result back to the model for a final explanation.

This separation is important. The model proposes the action, while application code remains responsible for authorization, validation, execution, error handling, and access to sensitive systems. Function calling can make an agent easier to integrate, but it does not remove the need for normal software safeguards.

01.AI’s documentation describes tool definitions, tool-choice controls, and structured JSON output for function invocation. The research confirms structured tool-call output, but it does not separately verify a legacy JSON mode or a JSON Schema-constrained-output feature. Those capabilities should not be assumed from the model’s function-calling support alone.

Context window and output limits

01.AI’s API documentation lists a 32K-token context window for Yi Large FC. The context window is the combined amount of text and other request information the model can process in one request, including the conversation, system instructions, tool descriptions, and tool results. A larger tool catalog can therefore consume part of the available context even before a user’s request is included.

The 32K value is the official figure retained for this model. A secondary OpenRouter listing reports a different 16,384-token context value, so users should distinguish third-party metadata from the provider’s documentation. The provider’s current limit should be confirmed if the precise context capacity is important to an application.

A maximum output-token limit was not verified in the supplied research. Developers should not assume that the full remaining context can be emitted as output. They should check the current API documentation or account-specific model response for applicable generation limits.

Pricing and API access

01.AI’s documented pricing is $3 per 1 million input tokens and $3 per 1 million output tokens. Input tokens include the request content and may also include system instructions, conversation history, tool descriptions, and returned tool results. Output tokens include the model’s generated response, including structured tool-call content.

The documented interface is compatible with the OpenAI chat-completions style and uses the endpoint https://api.01.ai/v1/chat/completions. The research also indicates streaming support in the model metadata. This can help applications display or process generated output incrementally, although the practical behavior of streamed tool-call arguments should be tested in the current API implementation.

Pricing and endpoint details are documented by 01.AI, but current live availability was not independently verified. Yi Large FC remains listed in the API documentation, while the relevant documentation appears older than some of 01.AI’s current corporate materials. Before building a production dependency, confirm that the model is enabled for the intended account, region, and API environment.

Capabilities and trade-offs

The model’s clearest strength is specialized tool use. It is intended to turn a natural-language request into a structured operation, which can reduce the amount of application-side routing logic needed for straightforward agent workflows. It is also suitable for applications where a model must decide between answering directly and invoking one of several available functions.

The supplied editorial assessment gives Yi Large FC a reasoning score of 7 out of 10, a coding score of 7 out of 10, a speed score of 7 out of 10, and a cost score of 5 out of 10. These are editorial evaluations, not benchmark results or scores published by 01.AI. They suggest a balanced model for ordinary reasoning, code-related tasks, and responsive tool orchestration, but they should not be treated as a guarantee of performance for a particular workload.

At $3 per million input and output tokens, Yi Large FC is not positioned as the lowest-cost possible model. Its price may be easier to justify when structured tool selection is central to the application, but a simpler classification or routing model may be more economical for high-volume tasks that do not require language-model reasoning. Conversely, applications requiring advanced reasoning guarantees may need a different model, since no such guarantee is established in the supplied documentation.

Supported inputs and outputs

The model’s structured output is particularly relevant to tool calls. That does not mean it offers every form of constrained generation or a separately documented general-purpose JSON mode. Those features should be checked independently if an application requires strict schema enforcement.

Best use cases

• Agent orchestration: Select among application tools and coordinate multi-step workflows.
• Business automation: Convert user requests into structured operations such as lookups, updates, or service requests.
• Internal enterprise assistants: Connect a text interface to approved company systems while keeping execution in application code.
• Developer applications: Add function calling to chat, coding, support, or workflow products through an OpenAI-compatible API pattern.
• Tool-aware customer support: Decide when to retrieve account or order information instead of answering from conversation context alone.

For these uses, the model should be paired with strict argument validation, permission checks, logging, and clear handling for cases where no tool is appropriate. The model’s output should be treated as a proposed action rather than an automatically trusted command.

When to choose Yi Large FC

Choose Yi Large FC when function calling is the main requirement and a text-based model with a 32K documented context window fits the application. It is especially relevant when the model must choose between several user-defined tools, produce structured arguments, and participate in an agent workflow without requiring native media understanding.

Consider another option when the application needs image, audio, or video input; native media generation; a clearly verified current availability commitment; or advanced reasoning capabilities that are not established for Yi Large FC. A lower-cost, simpler model may be preferable for deterministic routing or basic extraction, while a more reasoning-focused model may be a better fit for difficult planning and analysis. A model with separately documented JSON Schema constraints may also be more appropriate when strict output validation is a core requirement.

Availability and verification notes

The model was reported by OpenRouter as released on August 2, 2024, but that date is secondary metadata rather than a directly verified 01.AI release announcement in the supplied sources. 01.AI’s official documentation confirms the model ID, tool-use purpose, pricing, 32K context listing, structured output, and chat-completions endpoint.

What remains uncertain is the model’s current live serving status and several operational details, including a maximum output-token limit, fine-tuning, caching, batch access, and a distinct general-purpose JSON mode. These uncertainties do not change what Yi Large FC is designed to do, but they matter when selecting it for a production system. Confirm the current model list, account access, regional availability, quotas, and limits directly with 01.AI before deployment.