Models.dev releases open-source database of AI model specs, pricing, and capabilities

Models.dev addresses a critical gap in the AI ecosystem: the lack of a single, comprehensive database for model specifications, pricing, and capabilities. This open-source project, started as a community effort, stores data in TOML files organized by provider and model. Users can access the entire dataset via a REST API (curl https://models.dev/api.json) and retrieve provider logos as SVG files. The database includes detailed fields such as cost per million tokens (input, output, reasoning, cache), context limits, supported modalities (text, image, audio), reasoning capability, tool calling, and structured output support. It also tracks release dates, knowledge cutoffs, and whether weights are open. This makes it invaluable for developers and teams who need to compare models rapidly or integrate comparisons into their tooling.

Contributing to Models.dev is straightforward. New providers can be added by creating a folder in providers/ with a provider.toml file (e.g., name, npm package name, API base URL if using OpenAI-compatible endpoints). Models are defined as individual TOML files in the provider's models/ directory, with fields for name, attachment support, reasoning, tool_call, temperature, and more. For wrapper providers that mirror existing models, an extends mechanism is available to avoid duplication, with the caveat that it should not be used inside canonical provider directories like anthropic, openai, google, etc. Logos should be SVG with currentColor for flexibility. The project is actively maintained and welcomes community contributions to keep data accurate and up to date, making it a go-to resource for anyone working with multiple AI models.