Google Veo 3.1: picture and sound from one pass

This article is an overview based on vendor information and publicly available reporting, not a scored test of our own. Prices and features change particularly fast in this field; the vendor's own terms always take precedence. The tools we have used ourselves for weeks are scored in the AI video generators & voice AI category.

The difference: sound comes with it

Most video models deliver silent footage, with audio added afterwards. Veo generates picture and sound together, which defuses a stubborn problem: lip movements that do not match the speech.

On the available information Veo leads on lip sync and audio quality, with output up into 4K. For dialogue scenes that is the single most important point.

The model is reachable through Google's subscriptions and through the API, billed per second depending on the variant. In practice that means a 30-second clip has a price you can work out in advance.

Who it pays off for

For teams already working in the Google environment the path is short: the same access as for Gemini, the same billing. For everyone else it is worth comparing against the models in our portraits of Kling and Seedance, which set their own priorities on length and product stability.

If you only want a good voice over existing material, you do not need a video model at all: ElevenLabs does that more cheaply and with more control.

Limits and duties

Here too: short run times per generation, weaknesses on complex motion, and costs that rise with the number of attempts.

Legally, the labelling duty for synthetic content remains, as does the prohibition on recreating identifiable people without consent. The checkpoints are in our guide.

In short: the benchmark for lip sync and audio, and the obvious pick inside the Google environment. Per-second billing at least makes costing predictable.