Best of Open-Source AI
Apps, agents, models and the tools around them: what each one does, what it needs, where it breaks. No fluff, and paid never moves a rank.
155 projects in 16 categories, from GitHub data of 11 October 2026. Rankings update twice a day.
- Every project fact-checked on GitHub before it is listed
- Every entry has strengths and weaknesses
- Demo links so you can try before you deploy
Top 10 right now
Best of Open-Source AI is a ranked list of 155 open-source AI projects: chat apps and assistants, desktop apps, agent platforms and frameworks, workflow automation, RAG, memory, model serving, open models, training, image, video and voice, vector databases, gateways, observability, sandboxes, data tools, MCP servers and learning material. Each entry has strengths, weaknesses, specs and a score from public GitHub signals.
| Rank | Project | Adoption | Freshness | Maintenance | Easy to run | Agent-ready | Score |
|---|---|---|---|---|---|---|---|
| 1 | OmniRouteOpenAI-compatible gateway that routes requests across hundreds of AI providers | 94 | 100 | 100 | 67 | 70 | 86 out of 100 |
| 2 | LiteLLMOpenAI-format gateway and Python SDK for calling 100+ LLM providers | 89 | 100 | 83 | 83 | 30 | 84 out of 100 |
| 3 | nanobotSmall Python agent runtime with bundled WebUI, TUI and chat channels | 81 | 100 | 93 | 83 | 30 | 84 out of 100 |
| #4 | LocalAIOne OpenAI-compatible server for text, speech, image and video models | 83 | 100 | 89 | 67 | 70 | 82 out of 100 |
| #5 | MilvusDistributed vector database with dense, sparse and hybrid search at scale | 82 | 100 | 84 | 67 | 70 | 81 out of 100 |
| #6 | LobeHubAgent workspace with builder, groups, scheduling and 10,000+ MCP skills | 86 | 100 | 83 | 67 | 45 | 80 out of 100 |
| #7 | OpenClawPersonal assistant gateway that answers in Discord, Slack, WhatsApp and Telegram | 100 | 100 | 91 | 50 | 45 | 79 out of 100 |
| #8 | llama.cppC/C++ inference engine serving GGUF models over an OpenAI-compatible API | 94 | 100 | 89 | 50 | 70 | 79 out of 100 |
| #9 | hindsightAgent memory server with retain, recall and reflect operations | 81 | 100 | 93 | 50 | 85 | 78 out of 100 |
| #10 | LightpandaHeadless browser written in Zig for AI agents and scraping | 85 | 100 | 92 | 50 | 70 | 78 out of 100 |
Momentum, Verified build, Docs and Privacy are not measured yet; their weight goes to the signals shown. A dash means the signal is not scored for that kind of project. Hover a number for its rating in words.
Pick by what you need
- ChatGPT, but at homeChat UIs
- Run models on my own GPUModel serving
- An assistant in my Telegram or WhatsAppAssistants
- Chat with my own documentsRAG and knowledge
- Wire AI into the rest of my toolsWorkflow automation with AI
- One API key for every providerGateways
- Turn websites and PDFs into clean textData and scraping for AI
- See what my LLM app is actually doingObservability
- Let agents run code without wrecking my serverSandboxes
Assistants
Personal AI assistants you run yourself and talk to through chat apps, with memory and the ability to act.
| Rank | Project | Score |
|---|---|---|
| 1 | nanobotSmall Python agent runtime with bundled WebUI, TUI and chat channels | 84 out of 100 |
| 2 | OpenClawPersonal assistant gateway that answers in Discord, Slack, WhatsApp and Telegram | 79 out of 100 |
| 3 | Hermes AgentTerminal and chat-app agent that writes its own skills and remembers you | 75 out of 100 |
| #4 | ZeroClawSingle Rust binary agent runtime with 30+ channels and hardware access | 75 out of 100 |
| #5 | cowagentPersonal AI agent with memory, skills and chat channels like Telegram and Slack | 66 out of 100 |
Chat UIs
Web front-ends for local or API models, usually with user accounts, chat history and file upload.
| Rank | Project | Score |
|---|---|---|
| 1 | LobeHubAgent workspace with builder, groups, scheduling and 10,000+ MCP skills | 80 out of 100 |
| 2 | AnythingLLMDocument chat and agent app with built-in RAG, MCP and multi-user support | 77 out of 100 |
| 3 | Open WebUISelf-hosted chat UI for Ollama and OpenAI-compatible APIs with RBAC and RAG | 76 out of 100 |
| #4 | big-AGIMulti-model chat workspace with Beam side-by-side model comparison | 71 out of 100 |
| #5 | Hermes WebUIBrowser UI for Hermes Agent with sessions and file browser | 69 out of 100 |
Agent platforms
Visual or code-first builders for agents and workflows, with orchestration, tools and deployment.
| Rank | Project | Score |
|---|---|---|
| 1 | LangflowVisual flow builder that deploys agents as APIs or MCP servers | 73 out of 100 |
| 2 | DifyVisual LLM app platform with workflows, RAG pipeline, agents and APIs | 72 out of 100 |
| 3 | AutoGPTBlock-based builder for agents that run on demand, schedule or trigger | 70 out of 100 |
| #4 | SimWorkspace to build, deploy and monitor agents with 1,000+ integrations | 70 out of 100 |
| #5 | MulticaIssue board where AI coding agents take assignments like teammates | 68 out of 100 |
Workflow automation with AI
Workflow and integration platforms with AI steps or agent nodes, for wiring models into the rest of your tools.
| Rank | Project | Score |
|---|---|---|
| 1 | n8nVisual workflow automation with code steps, AI agent nodes and 1500+ integrations | 76 out of 100 |
| 2 | ActivepiecesSelf-hosted workflow automation with TypeScript integrations, alternative to Zapier | 65 out of 100 |
RAG and knowledge
Document Q&A, knowledge bases and enterprise search over your own files and data.
| Rank | Project | Score |
|---|---|---|
| 1 | LightRAGGraph-plus-vector RAG server with web UI and Ollama-compatible API | 76 out of 100 |
| 2 | Open NotebookSelf-hosted NotebookLM alternative with podcasts and 20+ model providers | 75 out of 100 |
| 3 | RAGFlowRAG engine with document parsing, agentic retrieval and citations | 72 out of 100 |
| #4 | WeKnoraEnterprise knowledge base combining RAG Q&A, agents and generated wikis | 68 out of 100 |
| #5 | SurfSenseOffline NotebookLM alternative that turns documents into decks, reports and podcasts | 68 out of 100 |
Search
Private search engines and AI answer engines that keep queries on your host.
| Rank | Project | Score |
|---|---|---|
| 1 | MorphicSearch engine that answers with citations and renders rich inline components | 70 out of 100 |
| 2 | Local Deep ResearchAgentic research assistant with local LLMs, SearXNG and encrypted libraries | 65 out of 100 |
| 3 | GPT ResearcherResearch agent that writes cited reports from web and local documents | 64 out of 100 |
| #4 | VaneSelf-hosted answer engine with cited sources over SearXNG | 54 out of 100 |
| #5 | MAESTROMulti-agent research platform that writes long reports from documents and web | 30 out of 100 |
Voice
Speech-to-text, text-to-speech, voice agents and meeting tools that run locally.
| Rank | Project | Score |
|---|---|---|
| 1 | VoiceboxLocal voice studio for cloning, TTS, dictation and agent speech | 72 out of 100 |
| 2 | Speech-to-SpeechModular voice-agent pipeline exposed through the OpenAI Realtime API | 67 out of 100 |
| 3 | Pocket TTS100M-parameter CPU text-to-speech with streaming and voice cloning | 64 out of 100 |
| #4 | Kokoro-FastAPIOpenAI-compatible Kokoro-82M speech API in CPU and GPU images | 64 out of 100 |
| #5 | IndexTTSZero-shot TTS with emotion, speed and pronunciation control | 61 out of 100 |
Image and video
Generation UIs and pipelines for images and video, usually around diffusion models.
| Rank | Project | Score |
|---|---|---|
| 1 | ComfyUINode-graph engine for diffusion image, video, audio and 3D models | 75 out of 100 |
| 2 | MoneyPrinterTurboGenerates short videos from a topic with script, footage, voice and subtitles | 69 out of 100 |
| 3 | InvokeAICanvas-first web UI for Stable Diffusion and Flux image generation | 62 out of 100 |
| #4 | AI ToolkitFine-tuning suite for image, video and audio diffusion models, with GUI and CLI | 54 out of 100 |
| #5 | Kohya's GUIGradio GUI and CLI for Kohya diffusion training scripts | 52 out of 100 |
Security
AI agents and tools for penetration testing, red-teaming and finding vulnerabilities in your own apps and models.
| Rank | Project | Score |
|---|---|---|
| 1 | StrixAutonomous AI pentesting agents that validate findings with working exploits | 69 out of 100 |
Memory
Long-term memory engines that store and retrieve facts for agents across sessions.
| Rank | Project | Score |
|---|---|---|
| 1 | hindsightAgent memory server with retain, recall and reflect operations | 78 out of 100 |
| 2 | MemPalaceLocal verbatim memory for coding agents on ChromaDB with 45 MCP tools | 77 out of 100 |
| 3 | agentmemoryPersistent memory server for coding agents, exposed over MCP and REST | 75 out of 100 |
| #4 | OpenVikingContext database exposing agent memory, knowledge and skills as a filesystem | 74 out of 100 |
| #5 | Mem0Memory layer for agents with a self-hosted server, SDKs and CLI | 73 out of 100 |
Model serving
Inference engines and model servers that expose local models over an API.
| Rank | Project | Score |
|---|---|---|
| 1 | LocalAIOne OpenAI-compatible server for text, speech, image and video models | 82 out of 100 |
| 2 | llama.cppC/C++ inference engine serving GGUF models over an OpenAI-compatible API | 79 out of 100 |
| 3 | vLLMHigh-throughput LLM serving engine with OpenAI and Anthropic APIs | 75 out of 100 |
| #4 | OllamaRuns open-weight models locally behind a CLI and REST API | 74 out of 100 |
| #5 | colibriC inference engine that runs huge MoE models by streaming experts from disk | 70 out of 100 |
Gateways
LLM gateways and proxies for routing, caching, rate limits and cost control across providers.
| Rank | Project | Score |
|---|---|---|
| 1 | OmniRouteOpenAI-compatible gateway that routes requests across hundreds of AI providers | 86 out of 100 |
| 2 | LiteLLMOpenAI-format gateway and Python SDK for calling 100+ LLM providers | 84 out of 100 |
| 3 | freellmapiOpenAI-compatible router that fails over across free LLM provider tiers | 76 out of 100 |
| #4 | ContextForge MCP GatewayRegistry and proxy federating MCP, A2A, REST and gRPC behind one endpoint | 68 out of 100 |
| #5 | HigressEnvoy-based API gateway with LLM proxy plugins and MCP server hosting | 61 out of 100 |
Observability
Tracing, evaluation and prompt management for LLM applications.
| Rank | Project | Score |
|---|---|---|
| 1 | LangfuseTracing, prompt management and evals for LLM apps on ClickHouse | 74 out of 100 |
| 2 | PhoenixLLM tracing, evals, datasets and prompt playground built on OpenTelemetry | 74 out of 100 |
| 3 | promptfooCLI for evaluating and red-teaming prompts, agents and RAG | 71 out of 100 |
| #4 | MLflowTracing, evals, prompt registry and AI gateway plus classic ML tracking | 70 out of 100 |
| #5 | OpikTrace, evaluate and monitor LLM apps and agents, Apache-2.0 end to end | 64 out of 100 |
Vector databases
Vector stores and hybrid search engines for embeddings.
| Rank | Project | Score |
|---|---|---|
| 1 | MilvusDistributed vector database with dense, sparse and hybrid search at scale | 81 out of 100 |
| 2 | MeilisearchRust search engine API with full-text, vector and hybrid search | 69 out of 100 |
| 3 | ChromaEmbedding database with a four-function API for Python and JavaScript | 64 out of 100 |
| #4 | QdrantRust vector database with payload filtering, REST and gRPC | 63 out of 100 |
| #5 | pgvectorPostgreSQL extension for vector similarity search with HNSW and IVFFlat | 59 out of 100 |
Sandboxes
Isolated runtimes where agents execute code, browse or use tools safely.
| Rank | Project | Score |
|---|---|---|
| 1 | LightpandaHeadless browser written in Zig for AI agents and scraping | 78 out of 100 |
| 2 | ObscuraRust headless browser with CDP, native rendering and stealth mode | 71 out of 100 |
| 3 | NemoClawNVIDIA reference stack running OpenClaw and Hermes inside OpenShell sandboxes | 71 out of 100 |
| #4 | OpenShellPolicy-enforced sandbox runtime for autonomous agents with credential brokering | 69 out of 100 |
| #5 | microsandboxLocal microVMs for untrusted code with fork, snapshot and SDKs | 61 out of 100 |
Data and scraping for AI
Tools that turn web pages, PDFs and documents into clean text or structured data that models can use.
| Rank | Project | Score |
|---|---|---|
| 1 | Crawl4AIPython crawler that turns pages into LLM-ready markdown, with a Docker API | 72 out of 100 |
| 2 | FirecrawlWeb scraping and crawling API that returns LLM-ready markdown | 68 out of 100 |
| 3 | Jina ReaderConverts any URL or search query into LLM-friendly markdown | 44 out of 100 |
Open-source alternatives to paid tools
The highest-scoring open-source projects in the category we compare each product in. A project marked "own claim" calls itself an alternative in its README.
Instead of ChatGPT
Compared in Chat UIs- LobeHubAgent workspace with builder, groups, scheduling and 10,000+ MCP skills80 out of 100
- AnythingLLMDocument chat and agent app with built-in RAG, MCP and multi-user support77 out of 100
- Open WebUISelf-hosted chat UI for Ollama and OpenAI-compatible APIs with RBAC and RAG76 out of 100
Instead of Perplexity
Compared in Search- MorphicSearch engine that answers with citations and renders rich inline components70 out of 100
- Local Deep ResearchAgentic research assistant with local LLMs, SearXNG and encrypted libraries65 out of 100
- GPT ResearcherResearch agent that writes cited reports from web and local documents64 out of 100
Instead of LM Studio
Compared in Model serving- LocalAIOne OpenAI-compatible server for text, speech, image and video models82 out of 100
- llama.cppC/C++ inference engine serving GGUF models over an OpenAI-compatible API79 out of 100
- vLLMHigh-throughput LLM serving engine with OpenAI and Anthropic APIs75 out of 100
Instead of Glean
Compared in RAG and knowledge and Search- LightRAGGraph-plus-vector RAG server with web UI and Ollama-compatible API76 out of 100
- Open NotebookSelf-hosted NotebookLM alternative with podcasts and 20+ model providers75 out of 100
- RAGFlowRAG engine with document parsing, agentic retrieval and citations72 out of 100
Instead of Copilot Studio
Compared in Agent platforms- LangflowVisual flow builder that deploys agents as APIs or MCP servers73 out of 100
- DifyVisual LLM app platform with workflows, RAG pipeline, agents and APIs72 out of 100
- AutoGPTBlock-based builder for agents that run on demand, schedule or trigger70 out of 100
Instead of NotebookLM
Compared in RAG and knowledge- LightRAGGraph-plus-vector RAG server with web UI and Ollama-compatible API76 out of 100
- 75 out of 100
- SurfSenseown claimOffline NotebookLM alternative that turns documents into decks, reports and podcasts68 out of 100
Instead of Zapier
Compared in Workflow automation with AI- n8nVisual workflow automation with code steps, AI agent nodes and 1500+ integrations76 out of 100
- ActivepiecesSelf-hosted workflow automation with TypeScript integrations, alternative to Zapier65 out of 100
Instead of OpenAI API
Compared in Model serving- LocalAIOne OpenAI-compatible server for text, speech, image and video models82 out of 100
- llama.cppC/C++ inference engine serving GGUF models over an OpenAI-compatible API79 out of 100
- vLLMHigh-throughput LLM serving engine with OpenAI and Anthropic APIs75 out of 100
Instead of OpenRouter
Compared in Gateways- OmniRouteOpenAI-compatible gateway that routes requests across hundreds of AI providers86 out of 100
- LiteLLMOpenAI-format gateway and Python SDK for calling 100+ LLM providers84 out of 100
- freellmapiOpenAI-compatible router that fails over across free LLM provider tiers76 out of 100
Instead of LangSmith
Compared in Observability- LangfuseTracing, prompt management and evals for LLM apps on ClickHouse74 out of 100
- PhoenixLLM tracing, evals, datasets and prompt playground built on OpenTelemetry74 out of 100
- promptfooCLI for evaluating and red-teaming prompts, agents and RAG71 out of 100
Instead of Pinecone
Compared in Vector databases- MilvusDistributed vector database with dense, sparse and hybrid search at scale81 out of 100
- MeilisearchRust search engine API with full-text, vector and hybrid search69 out of 100
- ChromaEmbedding database with a four-function API for Python and JavaScript64 out of 100
Instead of Midjourney
Compared in Image and video- ComfyUINode-graph engine for diffusion image, video, audio and 3D models75 out of 100
- MoneyPrinterTurboGenerates short videos from a topic with script, footage, voice and subtitles69 out of 100
- InvokeAICanvas-first web UI for Stable Diffusion and Flux image generation62 out of 100
Instead of ElevenLabs
Compared in Voice- 72 out of 100
- Speech-to-SpeechModular voice-agent pipeline exposed through the OpenAI Realtime API67 out of 100
- Pocket TTS100M-parameter CPU text-to-speech with streaming and voice cloning64 out of 100
Instead of Otter.ai
Compared in Voice- VoiceboxLocal voice studio for cloning, TTS, dictation and agent speech72 out of 100
- Speech-to-SpeechModular voice-agent pipeline exposed through the OpenAI Realtime API67 out of 100
- Pocket TTS100M-parameter CPU text-to-speech with streaming and voice cloning64 out of 100
Instead of Mem
Compared in Memory- hindsightAgent memory server with retain, recall and reflect operations78 out of 100
- MemPalaceLocal verbatim memory for coding agents on ChromaDB with 45 MCP tools77 out of 100
- agentmemoryPersistent memory server for coding agents, exposed over MCP and REST75 out of 100
Runs on what you have
Full ranking per groupFrom each project README: whether it needs a GPU, and whether it mentions Apple Silicon or arm64. Apps and tools you run, not libraries.
| What you have | Projects | Top picks |
|---|---|---|
| No GPU needed | 102 | OmniRoute, LiteLLM, nanobot, LobeHub, OpenClaw |
| A GPU helps, not required | 33 | LocalAI, Milvus, llama.cpp, Open WebUI, ComfyUI |
| Needs a GPU | 8 | KTransformers, GPUStack, AI Toolkit, LMDeploy, Kohya's GUI |
| A Mac or another arm64 machine | 57 | OmniRoute, nanobot, LocalAI, Milvus, llama.cpp |
How this list is ranked
On this list: open-source projects whose main job is AI, from apps you run to libraries you build with. Tools for coding with AI, starters and the stack you ship on live on the sister list, Best of Vibe Coding. The score comes from public GitHub signals with published weights; see the methodology. Looking for coding agents, starters and the stack you ship on? See Best of Vibe Coding.