Leaderboard
Tools ranked by GitHub stars; models ranked by measured tokens/sec where we have a benchmark.
Top tools by stars
| # | Tool | Category | Stars | License | Data |
|---|---|---|---|---|---|
| 1 | n8n | RAG & agent frameworks | ★ 196,151 | Sustainable-Use-License | ☁ optional |
| 2 | Ollama | Inference runtimes | ★ 175,976 | MIT | 🔒 local |
| 3 | AUTOMATIC1111 Stable Diffusion WebUI | Image generation | ★ 164,196 | AGPL-3.0 | 🔒 local |
| 4 | Langflow | RAG & agent frameworks | ★ 151,749 | MIT | ☁ optional |
| 5 | Dify | RAG & agent frameworks | ★ 148,579 | Apache-2.0 | ☁ optional |
| 6 | Open WebUI | Chat UIs | ★ 145,130 | BSD-3-Clause | ☁ optional |
| 7 | LangChain | RAG & agent frameworks | ★ 141,589 | MIT | ☁ optional |
| 8 | ComfyUI | Image generation | ★ 120,445 | GPL-3.0 | 🔒 local |
| 9 | llama.cpp | Inference runtimes | ★ 120,129 | MIT | 🔒 local |
| 10 | OpenAI Whisper | Speech (STT / TTS) | ★ 104,791 | MIT | 🔒 local |
| 11 | vLLM | Inference runtimes | ★ 86,043 | Apache-2.0 | 🔒 local |
| 12 | RAGFlow | RAG & agent frameworks | ★ 84,862 | Apache-2.0 | ☁ optional |
| 13 | Lobe Chat | Chat UIs | ★ 79,758 | LobeHub-Community | ☁ optional |
| 14 | GPT4All | Inference runtimes | ★ 77,391 | MIT | 🔒 local |
| 15 | AnythingLLM | Chat UIs | ★ 63,157 | MIT | ☁ optional |
| 16 | PrivateGPT | RAG & agent frameworks | ★ 57,330 | Apache-2.0 | 🔒 local |
| 17 | CrewAI | RAG & agent frameworks | ★ 55,382 | MIT | ☁ optional |
| 18 | Flowise | RAG & agent frameworks | ★ 54,543 | Apache-2.0 | ☁ optional |
| 19 | whisper.cpp | Speech (STT / TTS) | ★ 51,731 | MIT | 🔒 local |
| 20 | Fooocus | Image generation | ★ 50,989 | GPL-3.0 | 🔒 local |
| 21 | LlamaIndex | RAG & agent frameworks | ★ 50,795 | MIT | ☁ optional |
| 22 | LocalAI | Inference runtimes | ★ 47,497 | MIT | 🔒 local |
| 23 | Text Generation WebUI | Inference runtimes | ★ 47,443 | AGPL-3.0 | 🔒 local |
| 24 | Coqui TTS | Speech (STT / TTS) | ★ 45,744 | MPL-2.0 | 🔒 local |
| 25 | Jan | Inference runtimes | ★ 43,514 | AGPL-3.0 | 🔒 local |
| 26 | Chatbox | Chat UIs | ★ 40,977 | GPL-3.0 | ☁ optional |
| 27 | LibreChat | Chat UIs | ★ 40,618 | MIT | ☁ optional |
| 28 | Quivr | RAG & agent frameworks | ★ 39,204 | Apache-2.0 | ☁ optional |
| 29 | Bark | Speech (STT / TTS) | ★ 39,191 | MIT | 🔒 local |
| 30 | Diffusers | Image generation | ★ 34,046 | Apache-2.0 | 🔒 local |
| 31 | Onyx | RAG & agent frameworks | ★ 30,842 | MIT | ☁ optional |
| 32 | SillyTavern | Chat UIs | ★ 30,581 | AGPL-3.0 | ☁ optional |
| 33 | SGLang | Inference runtimes | ★ 30,212 | Apache-2.0 | 🔒 local |
| 34 | InvokeAI | Image generation | ★ 27,602 | Apache-2.0 | 🔒 local |
| 35 | MLX | Inference runtimes | ★ 27,526 | MIT | 🔒 local |
| 36 | Haystack | RAG & agent frameworks | ★ 25,873 | Apache-2.0 | ☁ optional |
| 37 | Kotaemon | RAG & agent frameworks | ★ 25,537 | Apache-2.0 | ☁ optional |
| 38 | llamafile | Inference runtimes | ★ 25,372 | Apache-2.0 | 🔒 local |
| 39 | faster-whisper | Speech (STT / TTS) | ★ 24,221 | MIT | 🔒 local |
| 40 | Letta | RAG & agent frameworks | ★ 23,749 | Apache-2.0 | ☁ optional |
Fastest local models · measured on Apple M4 Max · 36 GB
| # | Model | Params | Quant | Speed | VRAM |
|---|---|---|---|---|---|
| 1 | Llama 3.2 3B | 3.0B | Q4_K_M | 130.5 tok/s | 5.6 GB |
| 2 | Qwen2.5-Coder 14B | 14.0B | Q4_K_M | 37.7 tok/s | 14.2 GB |