Apps · Self-hosted assistant with its own engine

AnythingLLM

by Mintplex Labs · MIT · 67,500 GitHub stars

Local-first workspace for chatting with your documents and running agents. Ships with a built-in engine, so it works with no other setup.

macOSWindowsLinux

Get AnythingLLM

At a glance

Best for: Private document chat and agents with nothing to configure.
Inference engine: built-in engine (llama.cpp)
GPU support: NVIDIA CUDA, plus Vulkan on AMD and Intel
Models: Its built-in engine, any llama.cpp-compatible model, or a model from Ollama, LM Studio or LocalAI
Features: Agent workflows
MCP tools
Chat window
Platforms: Desktop app for Mac, Windows and Linux, or Docker. Includes a built-in local engine and also connects to Ollama, LM Studio, LocalAI and KoboldCPP.

Sources: anythingllm.com

Install

macOS

Download from anythingllm.com/download

Windows

Download from anythingllm.com/download

Linux

Download from anythingllm.com/download

AnythingLLM includes its own local model engine (built on llama.cpp), so you can start without Ollama. It can also use Ollama, LM Studio or another server if you prefer.

Alternatives

Pi Hermes Agent OpenClaw OpenCode Codex CLI OpenHands Zed Cline Open Interpreter Goose

Aider Tabby Crush Qwen Code Kilo Code Open WebUI Khoj

Letta

Which app should I use?

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.