Onyx is a self-hostable LLM application layer and chat interface designed for hosted environments. It brings retrieval-augmented generation (RAG), web search, deep research, and custom agents to enterprise teams through a feature-rich web interface with over 50 indexing connectors.
Project overview
The application provides a unified web interface for large language models (LLMs) with built-in support for RAG, AI agents, deep research, and over 50 indexing-based connectors.
Project type
AI Agent · RAG · Model Runtime
Use cases
Chat Assistants · Knowledge Q&A · Search & Research
Deployment
Refer to project documentation
License
License pending
Best for
Enterprise teams and developers seeking a self-hosted application layer with a chat interface for RAG and AI agents.
Users needing a hosted environment interface with over 50 indexing-based connectors.
Key capabilities
Provides hybrid index search and AI Agents for information retrieval.
Delivers in-depth reports using a multi-step research flow.
Enables building AI Agents with unique instructions, knowledge, and actions.
Browses the web for up-to-date information with support for multiple providers and crawlers.
Generates documents, graphics, and downloadable artifacts.
Lets agents interact with external applications via flexible Auth options.
Executes code in a sandbox to analyze data, render graphs, or modify files.
Allows chatting via text-to-speech and speech-to-text.
Limitations and risks
Onyx Lite mode lacks vector index, background containers, and performance optimizations.
Standard Onyx requires additional components and background containers which increase resource usage.
Extra enterprise features are paywalled under the Enterprise Edition.
Getting started
Run the installation script via curl and choose Onyx Lite mode for a lightweight installation. Setup difficulty is medium due to the multiple available configuration options.
After installation, access the Chat UI web interface to begin interacting with configured models and connectors.
Alternatives and comparisons
An end-to-end LLM-based system blueprint covering data generation, training, RAG, AWS deployment, monitoring, and evaluation.
A Ruby AI framework for Rails that supports multiple providers, Chats, Tools, Agents, RAG, and Workflows.
A multi-tenant AI agent gateway in Go offering PostgreSQL isolation, multi-provider LLM support, and messaging channel integrations.