LongCat Avatar 1.6 tested: 3-step local workflow makes 2-min lip-synced video 4x faster

CeFurkan · reddit · 2026-11-20

A Reddit user shares a local LongCat Avatar 1.6 workflow: a 3-step sampling pipeline replaces the official 8-step one, running 4x faster and generating a full 2-minute lip-synced animation in one pass, with audio via OmniVoice zero-shot voice cloning. They call it the best current audio+image lip-sync model and published a 4K tutorial with comparisons.

Original post →

More from Multimodal

Multimodal channel →