An open-source alternative to Claude (33k+ stars)!

Onyx is an open-source alternative to Claude, a popular language model, boasting over 33k stars on GitHub. This innovative AI tool serves as a knowledge and context layer for teams and AI agents, connecting to various applications to index and surface knowledge from over 50 sources. Onyx prioritizes data sovereignty by offering flexible self-hosted deployments, ensuring sensitive information remains secure.

Beyond search capabilities, Onyx enhances LLMs with advanced features such as web search, sandboxes, skills, and more. It creates a comprehensive representation of knowledge across all connected sources, ingesting data along with metadata, permissions, and other relevant details. This enables quick and efficient access to information for various use cases, offering a more reliable, low-latency, and cost-effective solution compared to traditional approaches.

Onyx supports all major LLM providers, whether self-hosted (like Ollama, LiteLLM, vLLM) or proprietary (Anthropic, OpenAI, Gemini). By connecting your organization's knowledge within an air-gapped, self-hosted deployment, Onyx prevents unauthorized access to sensitive information, both externally and internally. Users can choose their preferred embedding model and LLM provider, running them locally if desired.

Deployment options include Docker, Kubernetes, Helm/Terraform, and guides for major cloud providers. Onyx offers two deployment modes: standard and lite. The standard mode provides the complete feature set, recommended for serious users and larger teams, while the lite mode functions as a lightweight AI chat UI requiring less resources. Lite mode is ideal for testing the Onyx user interface or teams focusing solely on the Chat UI and Agents functionalities.

To explore Onyx for free without deploying, users can visit Onyx Cloud. With its versatile capabilities, Onyx caters to teams of all sizes, from individual users to extensive global enterprises. For those interested in contributing to the open-source project, a detailed Contribution Guide is available.