Installation Overview ~ AnythingLLM

Installation Overview

AnythingLLM Docker is both a single-user and multi-user application you can install on any webserver using docker and leverage local LLMs, RAG, and Agents with little to zero configuration and full privacy.

Self hosting AnythingLLM via Docker is very popular and can be done locally or on cloud servicers (aws, google cloud, railway etc..).

Docker vs Desktop Version

There are some distinct differences in functionality between each offering. Both are open source.

You want AnythingLLM Docker if...

You need an easy setup, but server-based service for AnythingLLM to use local LLMs, RAG, and Agents locally
You want to run an AnythingLLM instance that many people can use at the same time
You want to be able to share information with our users on your instance you invite
You need admin and rule-based access for workspaces and documents.
You will publish chat widgets to the public internet
You want to access AnythingLLM from the browser

You want AnythingLLM Desktop if...

You want a one-click installable app to use local LLMs, RAG, and Agents locally
You do not need multi-user support
Everything needs to stay only on your device
You do not need to "publish" anything to the public internet. Eg: Chat widget for website

The below table is a non-exhaustive list of features supported between platforms.

Feature	Available on Desktop	Available on Docker
Multi-user support	❌	✅
Emeddable chat widgets	❌	✅
One-click install	✅	❌
Private documents	✅	✅
Connect to any vector database	✅	✅
Use any LLM	✅	✅
Built-in embedding provider	✅	✅
Built-in LLM provider	✅	❌
White-labeling	❌	✅
Chat logs	✅	✅
Agent support	✅	✅
Agent skills	✅	✅
Third-party data connectors	✅	✅
Password protection	❌	✅
Invite new users to instance	❌	✅
Text splitting configuration	✅	✅
Whisper model support	✅	✅
Full developer API	✅	✅
User management	❌	✅
Workspace access management	❌	✅
Website scraping	✅	✅