veo

Google DeepMind unveils Veo, a text-to-video generation model

Generates 1080p video over a minute long from text and image prompts, watermarked with SynthID; launched in private preview via waitlist rather than public release.

Models & capabilities