The Nostrnet voice assistant, with Gemini Flash model, is live. It's extremely fast and ideal for this particular use case.

The next update will introduce models like Phi-3 and Llama 3, which will run locally even on phones using WebGPU.

Although they will be a bit slower, they offer complete privacy. Isn't that cool?

https://webcore.live
https://video.nostr.build/f6507088873c6f31dce9877827be381b668375ea44b9a6c2a2c72ee38d46c621.mp4