Dify
Self-Hosted AIOpen-source platform for building production LLM apps
Replaces OpenAI Assistants, Vertex AI Agent Builder
Best Multi-User Self-Hosted AI for self-hosting
The 17 best Multi-User self-hosted ai you can self-host, ranked by community traction.
RAGFlow
Self-Hosted AIRAG engine with deep document understanding
Replaces NotebookLM
AnythingLLM
Self-Hosted AIAll-in-one desktop and server RAG application
Replaces ChatGPT, NotebookLM
Open NotebookLM
Self-Hosted AISelf-hosted AI notebook for documents and audio overviews
Replaces Google NotebookLM
Langfuse
Self-Hosted AIOpen-source observability and analytics for LLM apps
Replaces LangSmith
Onyx
Self-Hosted AIAI assistant connected to your company's knowledge
Replaces Glean, ChatGPT Enterprise
Kotaemon
Self-Hosted AIClean RAG UI for chatting with your documents
Replaces NotebookLM
MaxKB
Self-Hosted AIKnowledge base question-answering system built on LLMs
Replaces Intercom Fin
Wren AI
Self-Hosted AIOpen-source GenBI agent for talking to your database
Replaces ThoughtSpot
Bisheng
Self-Hosted AIOpen LLM application platform for enterprise workflows
Replaces Dify
Local Deep Research
Self-Hosted AIAI-powered deep research tool with multi-source search (arXiv, PubMed,
GPUStack
Self-Hosted AIManage GPU clusters for running AI models
Replaces OpenAI API
Agenta
Self-Hosted AIOpen-source LLMOps platform for prompt and app management
Replaces LangSmith
GPTLink
Self-Hosted AIQuickly deploy your own ChatGPT-style web service
Replaces ChatGPT
BionicGPT
Self-Hosted AIOn-premise generative AI platform for teams
Replaces ChatGPT Enterprise
SecureAI Tools
Self-Hosted AIPrivate AI chat and document analysis you fully control
Replaces ChatGPT
AI Horde
Self-Hosted AICrowdsourced distributed cluster for AI generation
Replaces Midjourney
No apps match these filters.
At a glance
| App | Deploy effort | Difficulty | Min. RAM | License | ★ Stars |
|---|---|---|---|---|---|
| Dify | Under-an-hour setup | Medium | 4,096 MB | Apache-2.0 | 158,399 |
| RAGFlow | Under-an-hour setup | Medium | 8,192 MB | Apache-2.0 | 92,165 |
| AnythingLLM | Under-an-hour setup | Medium | 4,096 MB | MIT | 67,036 |
| Open NotebookLM | Under-an-hour setup | Medium | 2,048 MB | MIT | 40,157 |
| Langfuse | Under-an-hour setup | Medium | 2,048 MB | MIT | 35,795 |
| Onyx | Under-an-hour setup | Medium | 8,192 MB | MIT | 32,444 |
| Kotaemon | Under-an-hour setup | Medium | 4,096 MB | Apache-2.0 | 25,801 |
| MaxKB | Under-an-hour setup | Easy | 2,048 MB | GPL-3.0 | 22,973 |
| Wren AI | Under-an-hour setup | Medium | 4,096 MB | AGPL-3.0 | 17,865 |
| Bisheng | Under-an-hour setup | Hard | 4,096 MB | Apache-2.0 | 12,044 |
Quick picks
- Most adopted: Dify (158,399 GitHub stars).
- Lightest footprint: Local Deep Research (from 512 MB RAM, against 8,192 MB for the heaviest here).
- Licensing: 7 of 17 use MIT; check the rest individually if you plan to modify and redistribute.
Every option here is open-source and self-hostable, tagged Multi-User within the self-hosted ai category. Deploy effort is graded from published container recipes, not vendor claims — see the methodology.