Veo 3.1 is a Google DeepMind video model, one of the first to generate video together with sound. The model understands audio descriptions right in the prompt: a beach scene gets crashing waves and seagull cries, rain in the city gets the patter of drops, and a presenter in frame speaks the line you specify, in sync.
The service offers Veo 3.1 Fast — quicker and cheaper than the full version, with the same signature realism. It renders motion physics, lighting and materials convincingly, and handles people and complex scenes with confidence. Videos can be generated from text or from a photo; duration is 4, 6 or 8 seconds, with 720p or 1080p resolution at the same price.
Sound is optional: with it, a clip costs 66 credits for 4 seconds, without it — 44. Veo 3.1 is a solid choice for ads, voiced clips and scenes where realism matters.