Onyx
✓ Text source-checked Oct 23, 2026Open-source AI enterprise search platform that helps teams find information across all company data.
Source-checked against the vendor's site · rebuilt from the vendor's own pagesProfile updated Link checked Pricing captured How we rate
In short
Onyx is an open-source, self-hosted enterprise search platform that connects to 50+ data sources. It provides secure, permission-aware AI answers using your choice of LLM providers.
Onyx connects to enterprise knowledge to provide accurate, fast, and token-efficient context for humans and agents. Onyx runs entirely in the customer's cloud or on bare metal infrastructure.
Onyx uses over 50 connectors to sync tools continuously and index them with the permissions each source enforces. Users can swap model providers in one click, route teams to different models, or run open weights on their own GPUs.
The product provides a workspace interface with chat and search capabilities for accessing recent chats and documents. Onyx integrates with Slack channels to answer questions directly within the thread.
The Business plan is available for teams of any size at $20 per user per month with annual billing. Onyx is available as a desktop app, MCP server, Slackbot, and Chrome extension.
According to Onyx: Onyx connects to enterprise knowledge to provide accurate, Fast, and token-efficient context for humans and agents. Its pages list this strength: The code is open source and auditable end to end.
— The AI Tool Bible editorial team
Pros
- ✅ The code is open source and auditable end to end.
- ✅ The vendor says that documents and embeddings stay in the customer's infrastructure, and no training is performed on the data by anyone.
- ✅ Source permissions are enforced at query time.
- ✅ It offers over 50 connectors that sync continuously and index with source-enforced permissions.
- ✅ Model providers can be swapped in one click, and open weights can be run on customer GPUs.
Pricing (captured )
| Tier | Monthly | Per seat | Key features |
|---|---|---|---|
| Business | $20 | Yes |
|
| Enterprise | Contact us | — |
|
Prices change; we show the grid as captured on Aug 7, 2026. Check current pricing on the vendor's site →
Facts
- Platforms
- chrome-extension (not source-checked)
- API available
- Yes
- Free trial
- Yes
- Open source
- Yes
- Model used
- LLM-agnostic — routes to OpenAI (GPT-4o/GPT-5), Anthropic Claude, Google Gemini, Azure OpenAI, AWS Bedrock, or local Ollama/vLLM models (not source-checked)
- Hacker News
- 7 stories · top 3 points · last Apr 8, 2026 · checked Oct 17, 2026
- Earliest Wayback capture
- 2024-11-18 (earliest archived capture of the site, not a launch date)
Where this tool sits
Use cases
Frequently asked
- How much does Onyx cost?
- Onyx offers a free plan. The Business plan costs $20 per user per month with annual billing. This plan is available for teams of any size and includes access to the full platform features.
- Which AI models does Onyx support?
- Onyx is LLM-agnostic. It routes to OpenAI (GPT-4o/GPT-5), Anthropic Claude, Google Gemini, Azure OpenAI, AWS Bedrock, or local models via Ollama and vLLM. Users can swap providers in one click or run open weights on their own GPUs.
- How does Onyx handle data security and permissions?
- Onyx runs entirely in the customer's cloud or on bare metal infrastructure. It uses over 50 connectors to sync tools continuously and indexes them with the specific permissions each source enforces, ensuring secure access to company data.
- What integrations are available for Onyx?
- Onyx integrates with Slack channels to answer questions directly within threads. It is also available as a desktop app, MCP server, and Chrome extension. It connects to over 50 other tools to sync and index enterprise knowledge continuously.
- Is Onyx suitable for engineering and sales teams?
- Yes. Use cases include engineering productivity with full context, sales enablement with instant access to conversations and product updates, and support team assistance. It helps teams find information across all company data for accurate, fast responses.