Inference
CloudEasyUnavailable
Fastest and most powerful. Uses cloud-hosted models via API.
Ollama (Local)Advanced
Run models on your own machine via Ollama. Fast iteration for development. Requires Ollama running at localhost:11434.
On-Device (Browser)Experimental
Maximum privacy. Models run entirely in your browser using WebGPU. First load downloads the model (~2GB), then works offline.
Display
About
Multigrain — for your information
Created by Abe Rubenstein
Sparked by Tim Brown's Train On Your Groupchat