PocketAI — v1.0 · Windows
Run open-source language models directly on your Windows PC. No internet connection, no subscription, no data leaving your machine.
Built from scratch to run locally on consumer hardware.
Your conversations never touch a server. Zero telemetry, zero tracking, zero cloud.
Download a model once. Use it forever with no internet connection required.
Powered by llama.cpp — the fastest open-source local inference engine available.
Choose from Llama 3, Phi-3.5, Gemma 2, and Mistral 7B. Download inside the app.
Conversation history, streaming responses, and a customizable system prompt.
No subscription, no usage limits, no hidden fees. PocketAI is permanently free.
All models run locally in the GGUF format. Start small, go bigger when you're ready.
Grab the Windows EXE below. No installer — just double-click and run.
Open the Models tab and click Download. The file is saved to your PC and works offline from then on.
Load the model, switch to Chat, and start a conversation. Everything runs on your machine.