Run a big model?

Paste or drag a Hugging Face / Ollama link, or a .gguf file. I'll check whether you can download it and whether it fits. Nothing downloads until you say yes.

Connect your computer (Ollama / llama.cpp)

This website talks to the AI apps on your computer directly. Nothing goes through our server.

Ollama: allow this site, then restart Ollama:

Windows (PowerShell):

llama.cpp: start a server with any GGUF file:

llama-server -m model.gguf --port 8080

Use Chrome or Edge, and allow “local network access” if the browser asks. Safari blocks sites from reaching localhost.

No model selected

Pick a model and start chatting

Small models run right in your browser with WebLLM — no install. Already have Ollama or llama.cpp? Connect them. Huge models can stream from Hugging Face.

Found a model on huggingface.co or ollama.com? Drag its link onto this page.

Pick a model and start chatting

Download?

Settings