AnythingLLM

For anyone who wants to ask questions of their own documents with a language model, local or from a cloud provider, from the desktop app or a server.

MIT Reviewed on October 29, 2026

By Jorge Aguilar · Checked against the official documentation · Updated October 29, 2026

View on GitHub (opens in a new tab) Official documentation (opens in a new tab)

What it does.

chat with your documents (PDF, TXT, DOCX and others) and see the cited sources
connect local llama.cpp, Ollama or LM Studio models, or providers such as OpenAI, Anthropic or Gemini
use agents, a no-code agent builder and MCP compatibility
host several users with permissions (Docker version only)

Systems: Desktop app for Mac, Windows and Linux; server with Docker (amd64 and arm64 images); Android mobile app in a separate repo.

What you need.

RAM, minimum: 2 GB Whole service Container with the language model hosted elsewhere (in the cloud or on another server). On AWS/GCP/Azure the guide also says "at least 2 GB". If you turn on the built-in LLM you need more memory, depending on the model. source for this requirement (opens in a new tab)
CPU, minimum: 2 cores Whole service On x86 the CPU must have AVX2. source for this requirement (opens in a new tab)
Disk, minimum: 5 GB Whole service Container requirements table; it grows with your documents, vectors and models. source for this requirement (opens in a new tab)
Disk, recommended: 10 GB Whole service Cloud server: the Docker guide says "minimum 10 GB recommended". source for this requirement (opens in a new tab)
RAM, another documented figure: 16 GB Deployment example Desktop app with a local model: recommended value for a basic experience (the docs give no minimum). Figure from one specific installation (with extra services included), not a minimum for the program. The server calculator shows it, but does not add it up. source for this requirement (opens in a new tab)
CPU, another documented figure: 8 cores Deployment example Desktop app with a local model: recommended value (any 8-core CPU). source for this requirement (opens in a new tab)
Docker, plus yarn and node, per the repo's Docker guide. source for this requirement (opens in a new tab)
Access to an LLM, local or remote (the repo includes an optional local engine). source for this requirement (opens in a new tab)

Documentation version reviewed: AnythingLLM documentation (container and desktop; no version number on the page). Reviewed on October 29, 2026. The figures do not include: The local or built-in language model (if you turn it on): it needs memory of its own depending on the model, and the docs recommend a GPU.

Check whether your machine can handle it. It adds up its RAM, CPU and disk together with other repos.

How to install.

Desktop: download the app for your system at anythingllm.com/download.
Docker: download the image:
```
docker pull mintplexlabs/anythingllm
```
Start the container with the command from the Docker guide (it mounts a storage folder, publishes port 3001 and uses --cap-add SYS_ADMIN).

Commands taken from the official documentation. Read them before you run them: it is your machine.

Permissions and data.

The guide's Docker command adds --cap-add SYS_ADMIN to the container.
It ships with anonymous telemetry on by default; turn it off with DISABLE_TELEMETRY=true or under Privacy inside the app.

Limits and warnings.

Multi-user, the embeddable widget and branding themes are Docker-version only.
Even if you turn telemetry off, there are connections to cdn.anythingllm.com (models) and to GitHub (context cache files).
With an external LLM or embeddings, your text goes to that provider and its terms apply.
The Docker guide mentions that, if you use other services on localhost, you must use host.docker.internal.

Real cost and cloud.

Free (MIT); the real cost is your hardware or the cloud APIs you connect.
The native embedder and the LanceDB vector database are the defaults and work on your machine; the LLM can be local or remote, and privacy depends on that.
The README links a hosted instance as an alternative to self-hosting it.

License.

MIT Permissive open source

MIT: permissive license. You can use, modify and redistribute it, including in commercial products, as long as you keep the copyright notice.

This is a summary to orient you, not legal advice. For commercial use, read the original LICENSE file.

Where all of this comes from.

Checked against the official documentation on October 29, 2026; this is not a lab test. Nothing was installed or measured to write this guide. If something has changed, the project's documentation is the authority.

How I review each repo.

Similar repos.

See all repos