Veo 3.1 AI

Veo 3.1 is Google DeepMind's flagship AI video model, released in January 2026. It's the first mainstream AI video generator with true 4K output, featuring native audio with lip sync, reference-to-video, and an extend feature that reaches up to roughly 148 seconds.

Veo 3.1 Fast costs 54 credits per 8-second generation. The audio tier, which adds native audio generation with lip sync, costs 80 credits.

Veo 3.1 upgrades Veo 3 in four key ways: true 4K resolution (a mainstream first), noticeably better audio/video synchronization, reference-to-video support for character and style control, and the extend feature that chains clips up to ~148 seconds. Lip sync accuracy is also significantly improved over Veo 3.

Yes. Veo 3.1 is the first mainstream AI video model to offer genuine 4K output, delivering detail that holds up on large displays and in professional editing pipelines.

Reference-to-video lets you supply one or more reference images - a character, product, or location - and Veo 3.1 keeps them consistent in the generated video. Combined with extend, it's ideal for longer narrative projects.

A single Veo 3.1 generation is 8 seconds. Using the extend feature, you can chain generations into continuous sequences of up to approximately 148 seconds - over two minutes of coherent AI video.