Built on Open Source

Autonomous AI agent platform that chains LLM calls with tool use, memory, and planning. The original open-source attempt at general-purpose AI agents.

State-of-the-art machine learning for PyTorch, TensorFlow, and JAX. Thousands of pre-trained models for NLP, vision, audio, and multimodal tasks.

User-friendly AI interface supporting Ollama, OpenAI-compatible APIs, and local LLMs. Features RAG, web search, model management, and multi-user support.

The React framework for production. Hybrid static and server rendering, route handlers, middleware, TypeScript support, and optimized bundling.

Framework for building context-aware reasoning applications. Provides composable components for chains, agents, retrieval, and memory across LLM providers.

Beautifully designed, accessible, and customizable components built with Radix UI and Tailwind CSS. Copy-paste into your project, fully ownable.

Open-source LLM application development platform with visual workflow builder, RAG engine, agent framework, and model management for production AI apps.

High-throughput and memory-efficient inference and serving engine for large language models. PagedAttention, continuous batching, and tensor parallelism.

Next-generation Node.js and TypeScript ORM with auto-generated, type-safe query builder, visual data browser, and declarative schema migrations.

Data framework for LLM applications. Connect custom data sources to large language models with advanced indexing, retrieval, and query engine capabilities.

Deep learning toolkit for text-to-speech with pre-trained models in 1100+ languages. Supports voice cloning, fine-tuning, and multi-speaker synthesis.

Headless TypeScript ORM with zero dependencies. Type-safe SQL queries, automatic migrations, and a studio GUI. Supports PostgreSQL, MySQL, and SQLite.

High-performance build system for JavaScript and TypeScript monorepos. Remote caching, parallel execution, and incremental builds for large codebases.

Framework for orchestrating role-playing autonomous AI agents. Enables collaborative intelligence through structured agent teams with defined roles and goals.

Authentication for Next.js applications. Supports OAuth, Magic Links, Credentials, and database sessions with built-in CSRF protection.

High-performance vector similarity search engine and database written in Rust. Supports filtering, payloads, distributed deployment, and quantization.

End-to-end NLP framework for building production-ready RAG pipelines, semantic search, and question-answering systems with modular, pipeline-based architecture.

Open-source AI-native embedding database. Simple API for storing, searching, and filtering vector embeddings with automatic collection management.

Low-level, accessible, unstyled UI component primitives for building high-quality design systems and web applications in React.

Automatic speech recognition with word-level timestamps and speaker diarization. Built on Whisper with forced alignment and VAD for production accuracy.

Cloud-native vector database with built-in vectorizer modules, hybrid search, generative search, and multi-tenant architecture for production AI apps.

Open-source framework for building real-time, multimodal voice and video AI agents. Supports WebRTC, WebSocket transports, and pluggable AI services.

Simple, lightweight session-based authentication library. Framework-agnostic with adapters for popular databases and a minimal, understandable API.

Build real-time multimodal AI agents that see, hear, and speak. Server-side Python SDK with built-in turn detection and voice activity detection.