Lip-sync test
Two-person exchange
The prompt: “Two people at a table. One asks "Did you see the new model?" and the other replies "It is better."”
Turn-taking requires the non-speaking face to stay still and listening. Models often animate both mouths.
Every model, same prompt
#1
Grok Video
xAI
5.00/5 best on this prompt
Correct turn-taking, both speakers synced across the cut.
#3
Kling 3.0
Kuaishou
3.83/5
Correct turn-taking mouth movement between the two speakers, but silent — no audio track.
Ranked by our blind vision-model judge. Read how we test.
This page ranks by benchmark score only. See affiliate disclosure.