Kling 4.0 'more natural lip sync': what you can call on Sume today

Kling's Kling 4.0 page lists 'More natural lip sync, tighter A/V sync' as a feature, but the same page shows Kling 4.0 Flash in limited rollout and full Kling 4.0 coming in October, with no pricing (read 2026-10-02). Sume's OpenAPI lists no Kling 4.0 id, so today a talking face on Sume comes from a still plus audio, not from a Kling 4.0 clip.

What does Kling say about 4.0 lip sync?

The Kling 4.0 feature page says the new model has dual-channel stereo audio, dialogue in nine languages, and 'More natural lip sync, tighter A/V sync'. It gives 30 seconds for 4.0 and 20 seconds for 4.0 Flash, with Flash at 720p.

It does not publish an API price or a lip-sync accuracy number, so there is nothing to compare on cost or quality.

What does Sume use for a speaking face?

Sume's models overview says video models do not lip-sync to generated TTS or to a later voice-over, and that every on-camera speaking shot is Fabric with an accepted still and TTS. MiniMax H3 Max Lip Sync is the explicit alternative inside a 5 to 14.8 second audio window.

Both take one still (or a ready avatar) and an audio URL, and the output length follows the audio.

What should I do while Kling 4.0 is rolling out?

Build the speaking-shot step on Fabric or H3 Max Lip Sync now, since the request body is shared and you can switch between them. Revisit when Kling publishes an API page and Sume lists a matching id; check video-router models in the docs rather than assuming.

For the other Kling 4.0 questions, see Kling 4.0 API: what Sume lists.