Interactive characters reportedly use Fable 23B to respond to screen and voice context
A user says the model takes around 30 milliseconds to start streaming tokens, praising the characters’ realism, speed, and interactivity.
TLDR
A user describes interactive characters that see what’s on your screen, listen to you, and respond using that context. The post says they use Fable 23B, which takes around 30 milliseconds to start streaming tokens. The user traces the idea back to Microsoft Clippit but calls these characters more realistic, fast, and interactive than earlier iterations.
Combined views
14.3K
2 Sources, first seen 15d ago