Sync react-1 model_mode and emotion vs Sume TTS emotion and speed

Sync's react-1 takes a video plus audio and edits the face: options.model_mode picks lips, face (the default) or head, and options.prompt takes a single-word emotion. Sume has no such switch on the video; emotion and speed are set in the text-to-speech request, and the face is then driven by that audio.

react-1 facts are from Sync's React models page; Sume facts from the OpenAPI document and the Models docs, read 2026-10-01.

What do react-1's modes do?

What can the emotion prompt say?

Sync documents single-word emotions only: happy, angry, sad and neutral. Without a prompt the model follows the emotional context of the input video. The page says model_mode works only with react-1 and is ignored by other models, and Sync's free-trial page lists react-1 as paid plans only, with a 15 second input limit in its generation-times table.

Where do emotion and pace live on Sume?

In the TTS request. Its generation_config takes emotion ("Optional emotion guide for generation."), speed ("Speed multiplier in [0.6, 1.5].") and volume ("Volume multiplier in [0.5, 2.0]."). The Models page says every on-camera speaking shot is Fabric with an accepted still plus TTS audio, and video models do not lip-sync to generated TTS or a later voice-over. So you shape the performance in the audio, then pass that audio to the talking-video call. The emotion field is free text up to 64 characters, not a fixed list, and it is a guide, not a guarantee: listen to the result.

Is there a head-movement control?

Not in the docs I read. Sume's talking-video docs describe an avatar, a script or scenes, a quality tier and an aspect ratio, not a face or head mode. For related emotion and pacing advice see AI avatar emotion and speaking speed.