Build on open.
Own your stack.

The UI library that AI interfaces are built with. Server components, hooks, and the ecosystem that powers every major AI web app. Start here for AI-powered frontend.

Pioneering autonomous AI agent. Give it a goal, it plans and executes steps using web browsing, file operations, and LLM reasoning. The project that sparked the agent framework wave.

Run LLMs locally with a single command. Llama 4, Mistral, Phi, Gemma, and 100+ models. No cloud, no API keys, no latency bills. macOS, Linux, Windows.

Modern Python web framework with native async support. Auto-generates OpenAPI docs, type validation with Pydantic, and seamless LLM API integration. The standard for AI backend services.

Framework for building LLM applications. Chains, agents, memory, and tools. The original batteries-included approach to composing LLMs into real applications.

Workflow automation with 400+ integrations. AI-native with native LLM nodes, memory, and agent execution. Self-host on your own infra or use cloud.

ByteDance's open-source multi-agent framework. Coordinates specialized sub-agents, Docker/Kubernetes sandboxes, TIAMAT cloud memory, and built-in skills for research and automation.

Open-source AI agent platform. MCP-native, multi-model support, sandboxed code execution, and web auth. Built for production from day one.

Framework for building autonomous agent crews. Role-based agents collaborate on shared objectives. Clean API, strong observability, and production deployment ready.

Self-hosted AI coding agent. Runs as a CLI with optional desktop integrations. Full codebase awareness, agentic planning, and execution — your models, your infra.