Danswer
Danswer is an open-source, AI-powered assistant designed to streamline workplace search across an organization. It is designed to integrate...
Last verified:
What is Danswer?
Onyx (formerly Danswer) is an open-source AI assistant and enterprise search platform that connects to your company's documents, applications, and people to provide natural language Q&A and chat capabilities. It serves as a unified interface for Generative AI, allowing teams to ask questions in plain language and receive accurate answers grounded in their private knowledge sources. The platform combines hybrid search, advanced Retrieval Augmented Generation (RAG), contextual retrieval, and LLM-based knowledge graphs to deliver reliable responses with minimal hallucinations.
Key features include deep research capabilities, MCP (Model Context Protocol) support, code interpreter for complex calculations, web search integration, chat sharing, user feedback mechanisms, and usage analytics. Onyx supports 40+ app connectors including Slack, GitHub, Confluence, Google Drive, Jira, Notion, and more, with real-time sync and fine-grained access control inheritance. The platform offers custom AI agents with tailored instructions, Actions for automation via MCP/OpenAPI, developer APIs for flexible configuration, and both web app and Slack bot interfaces.
Onyx is designed for enterprises and teams of any size who need secure, private AI access to their internal knowledge. It's particularly valuable for engineering teams shipping faster with full context, sales teams closing deals with instant access to conversations and product updates, support teams answering questions across their entire product, and company-wide deployment to empower every member with secure GenAI access. The platform is model-agnostic, supporting any LLM from OpenAI to Claude to Llama, and can be self-hosted on any infrastructure for complete data control.
Danswer pricing
Pricing model: Free
Onyx offers two main pricing options. The Community Edition is completely free under the MIT license with all core features including chat, search, 40+ connectors, custom agents, code interpreter, web search, and developer APIs - ideal for self-hosted deployments. The Business plan costs $20 per user per month with annual billing, includes Chat and Search UI, access to all major LLMs, custom AI agents, Actions (MCP/OpenAPI), 40+ app connectors, web search, deep research, code interpreter, image generation, Slack integration, developer APIs, query history & usage dashboards, basic auth, Google OAuth, RBAC, permission inheritance, encryption of secrets, and community support. A free trial is available with no credit card required. The Enterprise plan has flexible custom pricing for large organizations and includes everything in Business plus OIDC/SAML SSO, on-premise deployments, region-specific deployments, early access to features, white-labelling, custom integrations, hook extensions, data exports, invoice billing, volume discounts, higher education pricing, dedicated support, and enterprise SLA.
Danswer pros
- Open source with MIT license for complete freedom
- Self-hostable on any infrastructure for data privacy
- Model-agnostic - supports any LLM provider
- 40+ pre-built app connectors with real-time sync
- Fine-grained access control inheritance from sources
- Advanced RAG with hybrid search and knowledge graphs
- Deep research capabilities for complex queries
- Built-in code interpreter for data analysis
- Web search integration for latest information
- Custom AI agents with custom instructions
- Actions via MCP and OpenAPI for automation
- Developer APIs for flexible configuration
- Chat sharing and collaboration features
- Usage analytics and query history dashboards
- Slack bot and Chrome extension available
- No per-user licensing fees for self-hosted CE
- Full code transparency and infinite extensibility
- Role-based access control (RBAC) support
- Single Sign-On (SSO) in Enterprise Edition
- 30x ROI demonstrated at enterprise customers
Danswer cons
- Self-hosting requires technical DevOps expertise
- Free Community Edition lacks SSO and advanced enterprise features
- Enterprise pricing is custom (not transparent)
- No official mobile app yet (coming soon)
- API availability is limited compared to some competitors
- Requires separate LLM provider setup and API keys
- Community support only for free tier
- Complex setup for multiple data source connectors
- Enterprise features require contacting sales
- Documentation can be overwhelming for beginners
- No free cloud tier - only free trial required
- Limited third-party integrations beyond 40+ connectors
- Warm-up time needed for indexing large document sets
- Advanced features like workflows still coming soon
- No built-in data export in Community Edition
Frequently asked questions about Danswer
What is Onyx and how is it different from ChatGPT?
Onyx is an open-source AI assistant specifically designed for enterprise use that connects to your company's private documents, apps, and people, whereas ChatGPT is a general-purpose AI without access to your internal knowledge. Onyx provides reliable responses grounded in your team's knowledge using advanced RAG, hybrid search, and knowledge graphs, scoring significantly better on 99 real workplace questions across 220K internal documents compared to ChatGPT, Claude, and Notion AI in blind tests. Onyx also offers data privacy through self-hosting options and model-agnostic support.
Is Onyx really free to use?
Yes, Onyx Community Edition is completely free under the MIT Expat license with ALL core features including chat, search, connectors, custom agents, code interpreter, and developer APIs. You can self-host it on your own infrastructure at no licensing cost. The only costs are your infrastructure (server/cloud) and LLM provider API fees. The Business cloud plan is $20/user/month for teams wanting managed hosting, and Enterprise has custom pricing for advanced features like SSO and on-premise deployments.
What apps and data sources does Onyx connect to?
Onyx supports 40+ app connectors including Slack, GitHub, Confluence, Google Drive, Jira, Notion, Zendesk, Hubspot, Salesforce, local and remote files, websites, and more. All connectors are plug-and-play, sync updates in real time, and respect fine-grained access controls from the original sources. You can see all available connectors through the 'See All Connectors' link on their website.
Can I use my own LLM provider with Onyx?
Yes, Onyx is completely model-agnostic and supports any LLM provider including OpenAI, Claude, Llama, and other open-source models. This is a core enterprise feature that prevents vendor lock-in and enables security/privacy by allowing companies to run open-source AI models on their own hardware without sharing data with third parties. You configure your LLM provider from the Admin page.
How does Onyx ensure answer reliability and reduce hallucinations?
Onyx uses a custom RAG pipeline combining hybrid search, advanced Retrieval Augmented Generation, contextual retrieval, and LLM-based knowledge graphs to ensure the most relevant results and least hallucinations. Every answer is backed by quotes and reference documents so you can always trust what you get back. In blind tests on 99 real workplace questions across 220K internal documents, scored by two independent LLM judges, Onyx outperformed ChatGPT, Claude, and Notion AI.
What security features does Onyx offer for enterprise use?
Onyx is built from the ground up with security as a core commitment. Features include self-hosting for complete data control, model-agnostic architecture (no third-party data sharing), fine-grained access control inheritance from connected sources, role-based access control (RBAC), encryption of secrets, Single Sign-On (SSO) with SAML and OIDC in Enterprise Edition, permission inheritance from connected sources, and compliance-ready architecture. Enterprise customers get dedicated support and enterprise SLA.
Can I deploy Onyx on my own infrastructure?
Yes, Onyx is designed for easy self-hosting on any infrastructure - from a laptop to on-premise servers to cloud deployments. It's delivered as a set of Docker containers that can be deployed anywhere, and all data is processed and persisted on your instance. The self-host guide is available in the documentation, and the Community Edition is free to deploy with all core features. Enterprise customers can also get region-specific and on-premise deployments.
What are custom AI Agents and how do I create them?
Custom AI Agents in Onyx are AI assistants tailored to your specific use cases with custom instructions, knowledge, and prompts. The platform includes a UI for configuring Personas (AI Assistants) and their Prompts. Agents can be created with specific knowledge bases, custom instructions for particular departments (engineering, sales, support), and integrated with Actions for automation. You can create agents for different team needs through the admin interface.
Does Onyx offer API access for developers?
Yes, Onyx supports a wide range of Developer APIs that provide flexible options like query expansion, agent behavior configs, and more. The APIs are model-agnostic and designed for extensibility. Onyx makes development easy with clean interfaces, a built-in agent framework, and developer guides. The platform is fueled by an open-source community of thousands and offers full transparency into the code with infinite extensibility.
What is the ROI of using Onyx for my team?
Onyx delivers 30x ROI based on enterprise customer results. At Ramp, Onyx answers thousands of questions per week with superior answer reliability compared to other AI tools. The pricing page estimates that for 500 users making 80 queries per day, Onyx saves 80 minutes per query and 5 minutes per search, resulting in $2,880,000 saved per year. This translates to 6 hours per working day recovered for your team, enabling significant productivity improvements across all departments.