AnythingLLM
A private AI workspace that chats with your documents and runs agents, with local or cloud LLMs, on desktop or Docker.
Desktop app or Docker server with built-in RAG, agents, MCP and multi-user permissions.
AnythingLLM is an all-in-one app for building a private ChatGPT over your own documents. Firms and solo professionals who want document Q&A with a choice of local or cloud models should look at it, especially if a team needs shared access.
What it is
You connect a local or cloud LLM, drop in PDFs, DOCX and TXT files, and chat with them with source citations. It also has a no-code agent builder, MCP compatibility, scheduled tasks, memories and a developer API, and it supports many vector databases with LanceDB as the default.
Who it's for
It suits small firms, consultants and teams who need an internal assistant over client or policy documents. Multi-user permissions, white-labelling and the embeddable website chat widget are Docker-only, so the desktop app alone will not serve a team.
Key features
Document chat with citations
Upload PDF, TXT, DOCX and other files and ask questions with source citations.
No-code agent builder
Build agent flows without code, plus custom agent skills and MCP-compatible tools.
Dynamic model routing
Rules you define send each chat to the provider and model that suits it.
Scheduled tasks
Run recurring prompts on a cron schedule with full agent capabilities.
Multi-user permissions
Control access and experience per user on a shared instance (Docker version only).
Embeddable chat widget
Put a custom chat widget on your website (Docker version only).
Up and running in 3 steps
Our honest verdict
What's great
- It connects to dozens of LLM providers and nine vector databases, so you are not tied to one model vendor.
- A built-in local engine runs models on CPU or GPU (NVIDIA via CUDA, AMD and Intel via Vulkan) without a separate install.
- Deployment templates exist for Docker, AWS, GCP, DigitalOcean, Render, Railway and others.
- The desktop app is a single download for macOS, Windows and Linux, so one person can start without servers.
Where it falls short
- Team features (multi-user, white-labelling, embed widget) need the Docker version, which means running and maintaining a server.
- Answer quality depends on the model and embedder you choose; local models need capable hardware.
- The app sends anonymous usage telemetry unless you set DISABLE_TELEMETRY or turn it off in the Privacy sidebar.
- It is a broad platform with many settings, so tuning retrieval for large document sets takes time.
Pricing & license
The software is free under the MIT licence, on desktop or self-hosted with Docker. The project's pricing page (https://anythingllm.com/pricing) shows hosted plans: Basic at $50/month, Pro at $99/month, and Enterprise on request; you bring your own LLM API key.
Open remote AI-training roles
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Is there a hosted version?
Yes. The pricing page lists a hosted Basic plan at $50/month and Pro at $99/month, plus a contact-us Enterprise tier. Self-hosting with Docker is free.
Can several people use one instance?
Yes, with the Docker version. Multi-user support and per-user permissions are listed as Docker-only features.
Does it send my data anywhere?
It collects anonymous usage events such as installation type and that a chat was sent, not their content. You can opt out with DISABLE_TELEMETRY, and external LLM providers you choose still receive your prompts.
