A self-hosted, provider-agnostic web interface for chatting with local Ollama models and OpenAI-compatible APIs, with RAG, plugins and multi-user access control.
onyx-dot-app/onyx
Self-hosted AI chat and knowledge platform that indexes data from 50+ connected apps and works with most LLM providers.
Open-source alternative to:ChatGPT Enterprise
Glean
- Stars
- 32,605
- Forks
- 4,583
- Open issues
- 591
- Last commit
- 11 hours ago
What is it?
Onyx is a knowledge and context layer for teams and AI agents. It connects to many business applications, pulls in documents along with their metadata and permissions, and builds an index so that answers can be retrieved from internal knowledge. Users can ask questions in a chat interface, and the platform supports RAG-based search, deep research reports, custom agents with specific knowledge subsets and instructions, and actions in external applications through MCP.
The platform can be deployed in a self-contained, air-gappable environment, with the document index, database, and processing kept inside the deployment. Embedding models and LLM providers can be self-hosted (such as Ollama, vLLM, or LiteLLM) or proprietary (such as OpenAI, Anthropic, or Gemini). Web search integrations, a secure code sandbox, artifact generation, and voice mode are also included.
Access is available through a web and desktop app, a Slack and Discord bot, an MCP server for clients like Claude Code, a Chrome extension, and an embeddable widget. Deployment options include Docker, Kubernetes, and Helm, and a lightweight Lite mode runs under 1GB of memory for chat and agents without document indexing.
Key features
- 01Agentic RAG search that combines a hybrid index with a retrieval-tuned agent
- 02Deep research workflow for multi-step, in-depth reports
- 03Custom agents with scoped knowledge, instructions, and actions
- 04Web search via Serper, Google PSE, Brave, SearXNG, and others
- 05External actions and MCP support for end-to-end tasks
- 06Enterprise features including SSO, RBAC, and query history
Use cases
- Giving a team a single chat interface to search across its internal tools and documents
- Building AI agents that answer questions from a specific set of company knowledge
- Running a private, self-hosted AI assistant in an air-gapped or sensitive environment
- Answering questions in Slack or Discord from organizational knowledge
Quick start
- 1
Run the official one-line install script from the README
- 2
Or try Onyx Cloud free without deploying
Things to know
- The standard deployment runs many components, including indexes, workers, inference servers, Redis and MinIO, so self-hosting takes real effort.
- Lite mode is lighter but cannot index documents.
- Enterprise Edition features are under a separate commercial license from the MIT-licensed Community Edition.
Topics
FAQ
- What is onyx?
- Self-hosted AI chat and knowledge platform that indexes data from 50+ connected apps and works with most LLM providers. Onyx is a knowledge and context layer for teams and AI agents. It connects to many business applications, pulls in documents along with their metadata and permissions, and builds an index so that answers can be retrieved from internal knowledge. Users can ask questions in a chat interface, and the platform supports RAG-based search, deep research reports, custom agents with specific knowledge subsets and instructions, and actions in external applications through MCP.
- Is onyx free to use? What license does it have?
- onyx is open source under the Other license. Check the license terms before using it in commercial products.
- Is onyx still maintained?
- Actively maintained. The last commit to onyx was on Oct 20, 2026.
- What can I use onyx for?
- Giving a team a single chat interface to search across its internal tools and documents Building AI agents that answer questions from a specific set of company knowledge Running a private, self-hosted AI assistant in an air-gapped or sensitive environment Answering questions in Slack or Discord from organizational knowledge
- How do I install onyx?
- Run the official one-line install script from the README Or try Onyx Cloud free without deploying
- Where can I get onyx?
- The source code and installation instructions are on GitHub: https://github.com/onyx-dot-app/onyx
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Memory operating system for LLMs and AI agents that stores, retrieves, and manages long-term memory as an editable graph.
Local-first AI memory system that stores conversation history verbatim and retrieves it with semantic search, exposed as an MCP server and CLI.
Agent memory system that stores, recalls and reflects on memories so AI agents learn over time instead of only replaying chat history.
Open-source personal AI assistant and Agent Harness that plans tasks, runs tools and skills, and keeps memory across web and IM channels.
An agent workspace for knowledge work and software development, offering a choice of models and open-source agent harnesses, runnable locally or hosted.