Models.dev: Open-Source Database of AI Model Specs, Pricing, and Capabilities

Models.dev is an open-source database of AI model specifications, pricing, and capabilities, maintained as a community-contributed project. The data is stored in the repo as TOML files, organized by provider and model, with logos as SVGs. It also powers an API and is used internally in opencode.

API Access

You can fetch all model data via a single API endpoint:

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

The Model ID field (e.g., openai/gpt-5) matches the identifier used by AI SDK. Provider logos are available as SVG files:

curl https://models.dev/logos/{provider}.svg

Replace {provider} with the Provider ID (e.g., anthropic, openai, google).

Contributing

To add a new model, first check if the provider already exists in the providers/ directory. If not, create a provider folder with a provider.toml file:

name = "Provider Name"
npm = "@ai-sdk/provider"  # AI SDK Package name
env = ["PROVIDER_API_KEY"]  # Environment Variable keys used for auth
doc = "https://example.com/docs/models"  # Link to provider's documentation

If the provider exposes an OpenAI-compatible endpoint, set npm = "@ai-sdk/openai-compatible" and include the base URL: api = "https://api.example.com/v1".

Optionally, add a logo.svg to the provider's directory using currentColor for fills/strokes. Then create a model definition TOML file in providers/{provider}/models/. Example model definition:

name = "Model Display Name"
attachment = true  # supports file attachments
reasoning = false   # supports chain-of-thought
tool_call = true    # supports tool calling
structured_output = true  # supports structured output
temperature = true   # supports temperature control
knowledge = "2024-04"  # knowledge-cutoff date
release_date = "2025-02-19"
last_updated = "2025-02-19"
open_weights = true   # weights publicly available

[cost]
input = 3.00          # per million input tokens (USD)
output = 15.00        # per million output tokens
reasoning = 15.00     # per million reasoning tokens
cache_read = 0.30     # per million cached read tokens
cache_write = 3.75    # per million cached write tokens
input_audio = 1.00    # per million audio input tokens
output_audio = 10.00  # per million audio output tokens
[limit]
context = 400_000     # max context window (tokens)
input = 272_000       # max input tokens
output = 8_192        # max output tokens
[modalities]
input = ["text", "image"]
output = ["text"]
[interleaved]
field = "reasoning_content"

For wrapper providers that mirror existing models, use extends (not detailed in the snippet).

Models.dev is an open-source, community-driven database. The source is on GitHub, and contributions are welcome to keep data up to date.