How many AI open-source libraries are there?

Thousands exist on GitHub, but we've curated the top 12 that actually ship in production. Categorized by use case.

Is this list updated?

Yes — every pick is reviewed weekly. Star counts refresh daily via the GitHub API. Last review: 22 May 2026.

Awesome · 12 curated picks

Awesome AI

Q: What is the best open-source AI library?

Our top pick is LangChain: Most mature LLM framework with broadest ecosystem.

Curated open-source AI tools for builders in 2026. Frameworks, vector DBs, local model serving, UI clients, and the AI engineering toolkit pros actually ship with.

Agent frameworks

LangChain

100k

langchain-ai/langchain

Most mature LLM framework with broadest ecosystem.

LlamaIndex

38k

run-llama/llama_index

RAG-first framework for document retrieval + generation.

CrewAI

24k

crewAIInc/crewAI

Role-based multi-agent crews, ships fast.

AutoGen

36k

microsoft/autogen

Multi-agent conversation framework by Microsoft Research.

Local LLM serving

Ollama

110k

ollama/ollama

Easiest way to run Llama 3.3, Qwen, DeepSeek locally.

Vector databases

Qdrant

22k

qdrant/qdrant

Rust-based production vector DB.

Chroma

16k

chroma-core/chroma

Simple Python-first vector DB.

pgvector

14k

pgvector/pgvector

Postgres extension for vector search.

Milvus

30k

milvus-io/milvus

Distributed billions-of-vectors DB.

Weaviate

12k

weaviate/weaviate

Hybrid search + GraphQL API.

AI chat UIs

OpenWebUI

65k

open-webui/open-webui

ChatGPT-clone UI for local Ollama.

LobeChat

50k

lobehub/lobe-chat

Premium-looking chat UI with plugins.

Want curator deep-takes on each pick?

Lifetime members get full ~100-word curator takes per repo with "use this if / skip if" clauses, plus 13 ready-to-ship stack bundles. ₹99 once.

See lifetime pricing

More awesome lists