An AI agent that lives in your pocket.
Search the web, read your documents, write files, run jobs on a schedule, and remember what matters. Powered by a model on your phone, or any model you choose. Your data never leaves the device.
68.6K+ downloads on Google Play
Privacy focused. Local First.
Everything an assistant should do, on your phone.
AnythingLLM Mobile ships with the tools that make any agent useful right out the box: web search and reading, document retrieval, file generation, scheduling, personalization, calendar, location and more. Small models pick the right tool for the job so they keep a workable context window without hogging memory.
Make AnythingLLM your phone's default assistant and long-press home in any app to bring it up. Ask about what's on screen, typed or spoken, and get answers from the model you picked.
Local first, model agnostic
Your phone is the agent. The model is up to you.
Download a small model from Hugging Face right from the app. AnythingLLM will display a fit badge tells you what your phone can actually run, and context windows scale to your device so replies start fast, even with no signal.
Need more than a phone can carry? Pair with LM Studio, Llama.cpp, Ollama or your AnythingLLM Desktop over your network, or plug in the cloud provider you already pay for. Same tools, same documents, same privacy. The tokens can come from anywhere; your data stays put.
Local and on your network
- On-device GGUF
- AnythingLLM Desktop
- LM Studio
- Ollama
- llama.cpp
- LocalAI
- LiteLLM
Cloud, when you want it
- Anthropic
- OpenAI
- Gemini
- OpenRouter
- AWS Bedrock
- DeepSeek
- xAI
- Together AI
- Fireworks
- Any OpenAI-compatible API
And it's already got what you want.
Chain-of-thought display, citations, fork and retry, auto-named threads, lock-screen notifications. The familiar AnythingLLM experience, sized for one handed autonomous productivity.
Sees what you see
Attach a photo and chat about it with a multimodal model, fully offline.
Talk to it
Tap the mic and speak. Native speech recognition keeps up with natural pauses.
Share anything into a chat
Photos, documents and links from any app. Links get scraped, documents get parsed.
Tuned to your device
Context windows scale to your RAM and prompts are laid out to keep the KV cache warm between turns.
Your chats, your files
Export any thread as PDF with images, Markdown, JSON or plain text.
Open source
GPLv3 on GitHub. Read the code, file an issue, or build it yourself.
Pick up where you left off on desktop
Pair with AnythingLLM Desktop or a self-hosted server on your network and chat with your full workspaces and document libraries from your phone.
Threads, tools and documents stay in sync across devices, with no account and nothing routed through a third party.
Got compute?
The same agent, on your desktop.
AnythingLLM Desktop runs larger local models with deeply embedded tools, agent flows and background jobs that feel native on Mac, Windows and Linux. Pair it with your phone and pick up any chat from either one.
Get AnythingLLM Desktop