Google Veo 3.1

What it is actually for

Veo 3.1 is the model you pick when you do not want to think about it. It is the closest thing the field has to a safe default, and the reason is one feature: it generates synchronised audio inside the same pass as the video — dialogue, ambient sound and music together.

That sounds like a nice-to-have until you build a pipeline. Every model without native audio forces a second generation step, which means another API call, another failure mode, more latency and a lip-sync problem you now own. Veo removes an entire stage from the workflow.

The trade-off

Price. Third-party sources put Veo 3.1 Standard at roughly $0.40 per second, several times what the value-tier models charge, with a Fast variant around $0.15. We have not verified either figure ourselves. For a marketing team producing a handful of finished spots that is irrelevant. For anything generating at volume it is the whole budget conversation.

Who should pick it

Teams producing realistic marketing and cinematic work where the output has to be client-ready, and anyone who needs enterprise procurement — through Vertex AI it comes with the SLAs and compliance paperwork that a startup API will not give you. It is also the default recommendation for anyone migrating off Sora, on feature parity grounds.

What we do not know yet

How often it needs a retry. Native audio raises the number of ways a generation can go wrong — a clip can be visually perfect and still unusable because the dialogue drifts. That failure rate is exactly what our benchmark measures and nobody publishes.