Conversational AI Open-Source Adoption by GitHub Stars (August 2026)
As of August 2026. Source: GitHub REST API. Stars measure developer mindshare, not usage or revenue. Method: scripts/collect_adoption.py.
| # | Project | Layer | GitHub stars | Forks | Language | Licence |
|---|---|---|---|---|---|---|
| 1 | ollama/ollama | Serving | 178,535 | 17,408 | Go | MIT |
| 2 | langflow-ai/langflow | Orchestration | 153,239 | 9,860 | Python | MIT |
| 3 | langgenius/dify | Orchestration | 152,436 | 24,068 | TypeScript | NOASSERTION |
| 4 | open-webui/open-webui | Interface | 148,767 | 21,661 | Python | NOASSERTION |
| 5 | langchain-ai/langchain | Orchestration | 144,246 | 24,018 | Python | MIT |
| 6 | ggml-org/llama.cpp | Serving | 123,903 | 21,704 | C++ | MIT |
| 7 | modelcontextprotocol/servers | Agents / Protocol | 89,561 | 11,445 | TypeScript | NOASSERTION |
| 8 | vllm-project/vllm | Serving | 89,044 | 20,682 | Python | Apache-2.0 |
| 9 | lobehub/lobe-chat | Interface | 81,690 | 15,808 | TypeScript | NOASSERTION |
| 10 | Mintplex-Labs/anything-llm | Interface | 64,705 | 7,129 | JavaScript | MIT |
| 11 | microsoft/autogen | Orchestration | 60,418 | 9,107 | Python | CC-BY-4.0 |
| 12 | crewAIInc/crewAI | Orchestration | 57,073 | 8,143 | Python | MIT |
| 13 | BerriAI/litellm | Orchestration | 56,332 | 10,575 | Python | NOASSERTION |
| 14 | FlowiseAI/Flowise | Orchestration | 55,362 | 24,898 | TypeScript | NOASSERTION |
| 15 | run-llama/llama_index | Orchestration | 51,637 | 7,928 | Python | MIT |
| 16 | mudler/LocalAI | Serving | 48,457 | 4,358 | Go | MIT |
| 17 | oobabooga/text-generation-webui | Interface | 47,544 | 5,978 | Python | AGPL-3.0 |
| 18 | milvus-io/milvus | RAG / Vector | 45,635 | 4,178 | Go | Apache-2.0 |
| 19 | danny-avila/LibreChat | Interface | 42,020 | 8,688 | TypeScript | MIT |
| 20 | chatwoot/chatwoot | CX / Automation | 35,859 | 8,610 | Ruby | NOASSERTION |
| 21 | n8n-io/n8n | CX / Automation | 34,953* | 60,131 | TypeScript | NOASSERTION |
| 22 | qdrant/qdrant | RAG / Vector | 33,971 | 2,576 | Rust | Apache-2.0 |
| 23 | chroma-core/chroma | RAG / Vector | 29,058 | 2,439 | Rust | Apache-2.0 |
| 24 | microsoft/semantic-kernel | Orchestration | 28,447 | 4,717 | C# | MIT |
| 25 | deepset-ai/haystack | Orchestration | 26,207 | 3,009 | Python | Apache-2.0 |
| 26 | modelcontextprotocol/python-sdk | Agents / Protocol | 24,002 | 3,784 | Python | MIT |
| 27 | pgvector/pgvector | RAG / Vector | 22,625 | 1,284 | C | NOASSERTION |
| 28 | RasaHQ/rasa | CX / Automation | 21,298 | 4,918 | Python | Apache-2.0 |
| 29 | pydantic/pydantic-ai | Orchestration | 19,292 | 2,516 | Python | MIT |
| 30 | weaviate/weaviate | RAG / Vector | 16,729 | 1,364 | Go | BSD-3-Clause |
| 31 | botpress/botpress | CX / Automation | 14,866 | 2,294 | TypeScript | MIT |
| 32 | livekit/agents | CX / Automation | 13,000 | 3,557 | Python | Apache-2.0 |
| 33 | huggingface/text-generation-inference | Serving | 10,886 | 1,277 | Python | Apache-2.0 |
| 34 | huggingface/chat-ui | Interface | 10,885 | 1,658 | TypeScript | Apache-2.0 |
* n8n is a general workflow-automation platform, not a conversational-AI tool. It is shown at its conversational-AI-adjusted figure (34,953): only about 17% of its 11,772 community workflow templates are conversational AI (1,293 chatbot + 758 RAG), and its raw star count is 200,617.