Overview

AnythingLLM is a fully open-source, all-in-one AI application that runs on your own computer or server. It wraps many of the most useful AI building blocks — chatting with your documents, research, retrieval-augmented generation (RAG), AI agents, and a developer API — into a single, friendly interface. Uniquely, it ships with a completely private default stack: everything from the model, the embedder, the vector database, and your chats and documents is stored and processed locally unless you choose to connect to an external provider. No accounts, no API keys, and no token limits are required to get started.

Built by Mintplex Labs and released under the MIT licence, AnythingLLM has grown a large community (64k+ GitHub stars and millions of Docker pulls). You can run it as a free desktop app on your own machine, or self-host the multi-user server with Docker to give your whole team a private AI workspace. It works with any LLM you like — local models through Ollama, LM Studio, or the bundled local provider, or enterprise models from OpenAI, Azure, AWS Bedrock, Anthropic, and others — so you stay in control of where your data goes.

Key Features

• Local-first by default: model, embedder, vector database, and storage all run locally for full privacy.
• RAG chat with your documents: ask questions over PDFs, Word docs, CSVs, and even entire codebases, with sources.
• Bring-your-own LLM: local models via Ollama/LM Studio or enterprise providers (OpenAI, Azure, AWS Bedrock, Anthropic).
• AI agents and custom tools: build workflows and custom tools without writing boilerplate.
• Multi-user self-hosting: deploy on your own server so a whole team can share a private workspace.
• Built-in developer API: use it as a local backend for your own applications.
• Multimodal support: handles text and image/audio-capable models in one interface.

Use Cases

• Private question-and-answer over your own documents and notes.
• A team knowledge base self-hosted on your infrastructure.
• On-device research and productivity assistant with no cloud dependency.
• An API backend for custom AI-powered tools and products.

Platforms

Windows, macOS, and Linux desktop; self-hosted via Docker.

Licence

MIT (open source)

Website

anythingllm.com · GitHub