DevHunt

sync-3 is a 16B parameter AI lip sync model that doesn't just move lips, it understands performances. Built on a global understanding of a person across an entire shot, it generates all frames at once instead of stitching isolated snippets. It handles what breaks every other model: close-ups, occlusions, extreme angles, low lighting - all while preserving the emotion of the original performance across 95+ languages in full 4K. Try it out at sync.so, via API, or in Adobe Premiere.

Screenshots

sync-3 screenshot 1
sync-3 screenshot 2
sync-3 screenshot 3
sync-3 screenshot 4
Watch Video
Added April 7, 2026View on Product Hunt

More in AI & LLM Tools

Feedback