Google’s new Omni family of generative models can turn images, audio, and text into short videos. In hands-on testing, The Verge found Omni Flash more capable than Google’s earlier Veo 3 at maintaining characters and accepting text-driven edits, though the outputs still show telltale AI glitches and inconsistency. For tech teams, the takeaway is that multimodal video generation is improving quickly and edging closer to production use cases, but quality and cost (credit-based pricing) still present real tradeoffs.