Tavus's Griffin starts talking before generation finishes, face reacts in 0.43s

TejasKumar_ · x · 2026-10-03

A developer tries out Tavus's new real-time avatar model Griffin, whose core breakthrough is breaking the legacy you-talk-then-I-talk relay of voice assistants:

The author notes it raises the question of humans' role once again — but very cool tech.

Original post →

More from Multimodal

Multimodal channel →