Oct 23, 2026
Block demonstrates Dessa, a full-duplex speech-to-speech model trained from scratch on human conversation without text or pretrained models, according to its research team. It can listen and speak at the same time.
The research preview explores speech-native conversation and expressive timing. Block reports local inference on a MacBook Pro M4 Max or better. The announcement offers selected safety researchers early access; it does not publish model weights or a generally available download.