AI Model Directory: The most comprehensive, auto-updating data layer for AI models.

Existing AI model catalogs are often opinionated, manual, or fragmented. We built the AI Model Directory to solve this: an open-source "source of truth" that provides standardized metadata for models across 50 providers (and counting). We've already accumulated 7000 models, which is more than models.dev, LiteLLM, or PortKey.

It's the same data engine that powers model selection in AgentOne, now available for everyone to use, fork, or contribute to.

Key Features:

• Truly Comprehensive: Tracks OpenAI, Anthropic, Google, DeepSeek, Mistral, and dozens of others.

• Always Up-to-Date: A GitHub Actions workflow runs every 24 hours to re-fetch metadata directly from provider APIs and documentation.

• Standardized Schema: Normalizes disparate data into a single schema covering pricing, token limits, modalities (text, image, audio, video), and features (tool calls, reasoning).

• Developer First: Data is stored as a tree of TOML files, so they're easy to read, diff, and consume. We also provide a bundled all.min.json for instant integration into your apps.

• Security Minded: Built with strict Zod schema validation, hardcoded endpoints, and sanitized outputs to ensure the pipeline is robust and safe.

How to use it:

• Browse: If you're not a developer, you can check out the searchable, sortable table at models.agent-one.dev. For example, you can use it to find free AI models across 50 providers.

• Integrate: Fetch the latest data directly from our GitHub raw URL to populate your model pickers or pricing tables.

• Contribute: Adding a new provider is usually just a single small adapter file.

Open source and free to use. Happy building! 🚀