Why it matters
Native audio and strong prompt adherence put it in the small group of models whose output needs no separate sound pass.
Veo 3.1
Added 2025-10-15Google's upgraded video generation model.

Veo is Google's video model, and like Sora its meaningful step is generating audio — including dialogue — together with the picture. Reference images let you steer look and subject rather than re-rolling until something matches. Access through Google's creative surfaces and API is the practical differentiator if your pipeline is already there. The constraints are the same across this whole class of tool: short durations, and continuity that needs planning at the shot level rather than the scene level.