Repolode
Submit
anomalyco

anomalyco/models.dev

An open-source, community-contributed database of AI model specifications, pricing, and capabilities, served via a public API.

View on GitHubWebsite
Stars
7,304
Forks
1,861
Open issues
385
Last commit
14 hours ago

What is it?

Models.dev collects specifications, pricing, and capabilities for AI models from many providers in one place. The data is stored as TOML files in the repository, organized by provider and model, and is used to generate the website and its API.

The project offers a JSON API at models.dev/api.json that returns provider endpoints and model details, with a provider-agnostic models.json and a combined catalog.json. Each model is keyed by an ID compatible with the AI SDK, and provider logos are available as SVG files by provider ID. Specialized model types such as decision models are excluded by default and can be requested with a type filter.

Contributors add providers and models through pull requests. Provider-specific pricing and limits can reuse shared model metadata through a base_model reference, and a GitHub Action validates submissions against the schema. The project is also used internally by opencode.

Key features

  • 01Provides a public JSON API covering AI model specs, pricing, and providers.
  • 02Offers provider-agnostic model metadata separately from provider-specific serving details.
  • 03Stores all data as TOML files organized by provider and model.
  • 04Supports filtering by model type via a query parameter.
  • 05Serves provider logos as SVG files.
  • 06Validates community contributions against a schema through a GitHub Action.

Use cases

  • Looking up context limits, modalities, or per-token pricing when choosing a model for an application.
  • Feeding a model list into tools or editors that need up-to-date provider and model data.
  • Attributing provider-specific model IDs to their originating lab through canonical_model_id.
  • Contributing new providers or model updates to keep the shared dataset current.

Quick start

  1. 1

    Query the full model catalog via the public API

    curl https://models.dev/api.json
  2. 2

    Filter the API to include specialized decision models

    curl "https://models.dev/api.json?type=decision"
  3. 3

    Fetch provider-agnostic model metadata only

    curl https://models.dev/models.json

Things to know

  • Specialized model types are omitted from the API by default, so you must pass a type parameter to see them.

Topics

#ai#llm#model-database#api#pricing#open-data#typescript

FAQ

What is models.dev?
An open-source, community-contributed database of AI model specifications, pricing, and capabilities, served via a public API. Models.dev collects specifications, pricing, and capabilities for AI models from many providers in one place. The data is stored as TOML files in the repository, organized by provider and model, and is used to generate the website and its API.
Is models.dev free to use? What license does it have?
models.dev is open source under the MIT license. Check the license terms before using it in commercial products.
Is models.dev still maintained?
Actively maintained. The last commit to models.dev was on Oct 19, 2026.
What can I use models.dev for?
Looking up context limits, modalities, or per-token pricing when choosing a model for an application. Feeding a model list into tools or editors that need up-to-date provider and model data. Attributing provider-specific model IDs to their originating lab through canonical_model_id. Contributing new providers or model updates to keep the shared dataset current.
How do I install models.dev?
Query the full model catalog via the public API: curl https://models.dev/api.json Filter the API to include specialized decision models: curl "https://models.dev/api.json?type=decision" Fetch provider-agnostic model metadata only: curl https://models.dev/models.json
Where can I get models.dev?
The source code and installation instructions are on GitHub: https://github.com/anomalyco/models.dev

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects