Onyx
Onyx is an open-source AI platform that connects to your enterprise knowledge, providing accurate, fast, and token-efficient context for humans and AI agents. It supports self-hosted deployment, 50+ connectors, and works with every major
Visit onyx-dot-app/onyxOverview
Onyx is an open-source knowledge and context layer that connects to 50+ enterprise applications, indexing and surfacing knowledge while protecting data sovereignty through self-hosted deployments. It provides accurate, fast, and token-efficient context for both humans and AI agents.
Key Features
- Agentic RAG with best-in-class search and answer quality based on a hybrid index and custom agent harness.
- Deep Research for multi-step research flows and in-depth reports.
- Custom Agents with unique knowledge subsets, custom instructions, and the ability to take actions.
- Web Search augmentation with live web search, supporting Serper, Google PSE, Brave, SearXNG, and more.
- External Actions & MCP integration to complete tasks end to end.
- Secure Sandbox for code execution and intermediate artifacts in complex workflows.
Use Cases
- Empower every member with secure access to GenAI and knowledge across the company.
- Ship faster with Generative AI and full context for engineering teams.
- Close more deals with instant access to every conversation and product update for sales.
- Answer questions confidently across your entire product for support teams.
Getting Started
- Deploy with a single command: curl -fsSL https://onyx\.app/install\_onyx\.sh | bash, or try Onyx Cloud for free without deploying.
Deployment & Requirements
- Supports Docker, Kubernetes, Helm/Terraform, and major cloud providers. Two modes: Standard (full feature set with vector index, background workers, AI inference servers, Redis, MinIO) and Lite (lightweight AI Chat UI under 1GB memory, no document indexing).
Before You Adopt
- License: Other. Review its terms before using, modifying, or distributing the project.
- Data sovereignty: Onyx can be run fully air-gapped with no external third parties, ensuring documents and embeddings stay in your infrastructure.
- Model provider flexibility: You can choose trusted embedding and LLM providers, including running them locally, to control external dependencies.
- Permission enforcement: Source permissions are enforced at query time to prevent internal data leaks.