Skip to content
AI for Mortals.

AnythingLLM

A private desktop app to chat with your documents using local or cloud AI models

Category
Chat
Last change
Oct 26, 2026
Try AnythingLLM

Changelog

October 2026

  1. New feature

    AnythingLLM Docker now includes a built-in local LLM powered by Lemonade and llama.cpp, with GPU passthrough, so no separate inference server is needed.

    What this means for you: AnythingLLM in Docker (a way of running software in a self-contained package) can now run an AI model by itself. You no longer need a separate program such as Ollama or LM Studio beside it. It uses your graphics card (GPU) if you pass one in, and recommends three models for your hardware.

    How to use it: It is off by default. Start the container with -e ENABLE_BUILT_IN_LLM=true, and add --gpus all for GPU speed-up (or pass in your AMD or Intel GPU). Built-in then becomes the default provider during onboarding. It runs on amd64 and arm64 machines. This is in v1.19.0.

    I want to run AnythingLLM in Docker with its new built-in local model, turned on with ENABLE_BUILT_IN_LLM=true, so my documents stay on my computer. Walk me through the steps in plain words. My computer is [Windows/Mac/Linux] with [GPU or no GPU].

    github.com

September 2026

  1. New feature

    AnythingLLM v1.16.2 adds Google Vertex AI as an LLM provider, a generate-image agent skill, You.com as default web search, and Hindi, Bulgarian and Swedish support.

    github.com

August 2026

  1. New feature

    AnythingLLM v1.16.1 overhauls the interface so multiple agent thinking and tool calls collapse into one element, and adds a navigation warning during inference.

    github.com

AI FOR MORTALS.

Your daily AI edge that will make everyone wonder how you do it all.

Every weekday, the practical AI updates that matter, and exactly what to do with them.

The latest in AI, in plain English. Free, every weekday.