Docs

Query API docs

The public API is free and requires no key. Use it to look up models, providers, labs, and rankings. Append .md to models, providers, labs, rankings, or docs addresses to get the Markdown text.

Plain text

A free query API

Every endpoint is open—no signup or API key. Use it to compare model capabilities, provider supply, and rankings. Cache results in your app and stay within the fair-use limit of 120 requests per minute.

Conventions

Path: /api/v1
Format: JSON, UTF-8
CORS: enabled for browser requests
Pagination: page, page_size (max 1000)

Markdown text

Append .md to models, providers, labs, rankings, or docs addresses to get the Markdown text for programs and AI. Addresses without a language prefix follow your language preference.

Docs/en/docs.md

Models/en/models.md

Providers/en/providers.md

Labs/en/labs.md

Rankings/en/rank.md

Endpoints

GET /api/v1/models

Search the model catalog by keyword, lab, provider, capability, input/output type, and sort order.

Parameters: q, lab, provider, reasoning, tool_call, structured_output, open_weights, attachment, modality, input_modality, output_modality, sort, order, page, page_size

Each item includes inputModalities, outputModalities, and the same values as modalities.input / modalities.output. Values include text, image, audio, video, and pdf. modality matches input or output; input_modality matches input only; output_modality matches output only. Comma-separated values must all be present.

GET /api/v1/models/{id}

Read a single model, its input/output types, and its pricing, limits, types, and thinking controls across providers.

Parameters: id uses lab/model, for example openai/gpt-5.5

Offerings include modalities.input and modalities.output for that provider. Values include text, image, audio, video, and pdf. Reasoning offerings include reasoningOptions. toggle means thinking can be turned on or off; effort.values lists intensity levels (none, minimal, low, medium, high, xhigh, max, default); budget_tokens is a thinking-token budget. An empty array means the model thinks, but callers cannot adjust it.

GET /api/v1/providers

List all providers and how many models they offer.

Parameters: q

GET /api/v1/providers/{id}

Read a provider and the models it lists, including pricing, limits, types, and thinking controls.

GET /api/v1/labs

List labs with their models and provider coverage. Descriptions are available in Chinese and English.

Parameters: q

GET /api/v1/labs/{id}

Read a lab and the models it publishes. Descriptions are available in Chinese and English.

GET /api/v1/rank

Rankings for overall, intelligence, or coding ability. Use q to filter by model, lab, or ID.

Parameters: q, category=overall|intelligence|coding, limit (max 1000)

GET /api/v1/coding-agents

Coding agent rankings by overall index, benchmark, cost, execution time, or token usage. Use q and model to filter.

Try it

Click “Run” to see the response.