Self-hosted alternatives to ChatGPT
Every prompt is a data point on someone else's balance sheet, and the model changes under you without notice.
Below are 10 open-source projects that cover the same ground, each scored 0–100 on how actively it is actually maintained — commit volume, release cadence and recent trend. 10 of the 10 are currently rated healthy or better. Ordering reflects that score, not sponsorship.
| Project | Health | Activity (11 mo) | Stars | Licence | Last release |
|---|---|---|---|---|---|
| LobeHub Modern design AI chat framework supporting multiple AI providers, one click install MCP Marketplace and Artifa | Thriving · 99 | 83k | ⊘ Proprietary | 27 Sept 2026 | |
| Open-WebUI User-friendly AI Interface, supports Ollama, OpenAI API. | Thriving · 98 | 153k | BSD-3-Clause | 21 Sept 2026 | |
| LocalAI Run your AI models locally and generate images and audio (alternative to OpenAI and Claude). | Thriving · 97 | 49k | MIT | 17 Sept 2026 | |
| LibreChat Enhanced ChatGPT-compatible AI chat interface supporting multiple AI providers, with multi-user auth, message | Thriving · 94 | 45k | MIT | 23 Sept 2026 | |
| Onyx Community Edition Chat UI that works with any LLM. It comes loaded with advanced features like agents, web search, RAG, MCP, dee | Thriving · 94 | 32k | MIT | 27 Sept 2026 | |
| Agenta LLMOps platform for prompt management, LLM evaluation, and observability. Build, evaluate, and monitor product | Thriving · 93 | 4.8k | MIT | 27 Sept 2026 | |
| AnythingLLM All-in-one desktop & Docker AI application with built-in RAG, AI agents, No-code agent builder, MCP compatibil | Thriving · 90 | 67k | MIT | 22 Sept 2026 | |
| Ollama Get up and running with Llama 3.3, DeepSeek-R1, Phi-4, Gemma 3, and other large language models. | Thriving · 89 | 182k | MIT | 25 Sept 2026 | |
| Local Deep Research AI-powered deep research tool with multi-source search (arXiv, PubMed, web), PDF text extraction, and encrypte | Thriving · 87 | 9.1k | MIT | 28 Aug 2026 | |
| LLM Harbor Containerized LLM toolkit. Run LLM backends, APIs, frontends, and additional services via a concise CLI. | Thriving · 85 | 3.2k | Apache-2.0 | 25 Sept 2026 |
How these were picked
Every project listed here is open source and can run entirely on hardware you control. We start from the categories that match ChatGPT's job — Generative Artificial Intelligence (GenAI) — then rank by health score so that a popular-but-abandoned project never appears above a smaller one that is actively shipping. The scoring method is documented in full.
The scores are recomputed from repository data on every rebuild, so this page reflects the state of these projects as of 1 Oct 2026 — not the day it was written.
LobeHub
Modern design AI chat framework supporting multiple AI providers, one click install MCP Marketplace and Artifacts / Thinking.
Best fit: Teams or individuals looking for a customisable, self-hosted web interface to interact with various generative AI APIs in one place.
Worth knowing: The software uses a proprietary licence, which may restrict how you can modify or redistribute the source code compared to standard open-source alternatives.
Open-WebUI
User-friendly AI Interface, supports Ollama, OpenAI API.
Best fit: It suits administrators who want a self-hosted graphical interface for local AI models without relying on third-party web services.
Worth knowing: Running local models requires significant hardware resources, and managing multiple backend integrations can introduce unexpected configuration overhead.
LocalAI
Run your AI models locally and generate images and audio (alternative to OpenAI and Claude).
Best fit: It suits developers and organisations who want to run generative AI workloads locally to maintain data privacy or bypass external service costs.
Worth knowing: You will need to provide your own compatible hardware, such as GPUs, to achieve acceptable inference speeds. Running complex models locally can also introduce significant resource management overhead.
LibreChat
Enhanced ChatGPT-compatible AI chat interface supporting multiple AI providers, with multi-user auth, message search, and plugin support.
Best fit: It suits administrators who want to host a centralised, multi-user AI chat interface with support for multiple upstream model providers.
Worth knowing: Managing API keys, user permissions, and upstream rate limits across multiple providers introduces ongoing operational overhead. It also duplicates the functionality of various native vendor applications.
Onyx Community Edition
Chat UI that works with any LLM. It comes loaded with advanced features like agents, web search, RAG, MCP, deep research, Connectors to 40+ knowledge sources, and more.
Best fit: Self-hosters who want a feature-rich, unified chat front-end to interface with multiple external or local language models under an MIT licence.
Worth knowing: Integrating numerous knowledge connectors and running advanced agentic workflows can introduce significant configuration overhead and high resource consumption.