Onyx
A self-hosted AI assistant and search for teams that connects to more than 50 workplace apps and answers from company knowledge while respecting permissions.
What is Onyx?
Onyx, formerly Danswer, indexes knowledge from tools such as Slack, Google Drive, Confluence, Jira and GitHub, together with their metadata and access permissions. People then ask questions in a chat interface and get answers grounded in documents they are allowed to see. It works with any model provider, adds web search, agents, skills and sandboxes, and can be deployed on your own infrastructure for data sovereignty. It offers 9 interface languages.
Good fit if
- Your company knowledge is spread across many SaaS tools and you want one AI search over it.
- Permissions matter: answers should only use documents a person can open.
Look elsewhere if
- You want document chat for an individual or a small group: AnythingLLM.
- You need a smaller deployment: the full compose stack runs about 14 services.
Before you adopt it
Onyx is MIT-licensed except the ee directories, which fall under a separate enterprise license for paid features. A Helm chart is included, and production setups use PostgreSQL, Redis, MinIO and a search index.
Drafted with AI from the project's README and repository data, then checked against that data on 30 October 2026. Report a correction
Onyx health: is it still maintained?
77 of 100, actively developed
Read from GitHub on 1 November 2026. How the score works
Onyx license: is it open source?
Onyx uses MIT, an open-source license approved by the OSI. Some features sit under a separate paid license inside the same repository.
- April 2023First published on GitHub.
- TodayMIT, with paid parts: license file assigns a different licence to part of the code; directory backend/ee/; directory web/src/app/ee/; directory web/src/ee/; separate license backend/ee/LICENSE.
How to run Onyx yourself
- Docker
- Dockerfile or compose file in the repository
- Compose file
- deployment/docker_compose/docker-compose.yml: 15 services (postgres, elasticsearch, proxy, redis, minio)
- Setup effort
- 5 of 5, judged from the compose file
- Kubernetes
- Helm chart in the repository
- Desktop and mobile
- Builds for Linux, Windows, macOS, including ARM in GitHub releases
- Interface languages
- 9 languages
Install instructions: onyx.app or the README on GitHub.
Alternatives to Onyx
- DocsGPTPrivate AI platform for agents, assistants and enterprise search. Built-in Agent Builder, Deep research,… 84 health · MIT · 18.4k stars
- Open WebUIUser-friendly AI Interface (Supports Ollama, OpenAI API, ...) 76 health · Open WebUI License · 157k stars
- FastGPTFastGPT is a knowledge-based platform built on the LLMs, offers a comprehensive suite of out-of-the-box… 82 health · FastGPT Open Source License · 30k stars
- HarborStop configuring your AI stack. Start using it. One command brings a complete pre-wired LLM stack with… 80 health · Apache-2.0 · 3.3k stars
- Hermes AgentThe agent that grows with you 76 health · MIT · 259k stars
Questions about Onyx
Is Onyx open source?
Yes. Onyx is published under MIT, an OSI-approved open-source license. Some features are kept under a separate paid license in the same repository.
Is Onyx still maintained?
Its health score is 77 of 100 (actively developed). It had 1,871 commits in the last 90 days, the latest on 1 November 2026, and its latest release came out on 30 October 2026.
How do I run Onyx?
You can run it with Docker, on Kubernetes with the Helm chart in the repository, as an app for Linux, Windows, macOS, including ARM. The README on GitHub has the exact steps.
What are the alternatives to Onyx?
Close alternatives in the AI chat & local LLMs category include DocsGPT, Open WebUI, FastGPT.