Google launches Gemini Omni Flash and Nano Banana 2 Lite

Google rolled out two new Gemini models today, and the more interesting story is the prices on the sticker, not the demo reel. According to Google's announcement, Gemini Omni Flash is priced at $0.10 per second of video output and currently offers 10-second video generations, while Nano Banana 2 Lite delivers text-to-image outputs in 4 seconds at $0.034 per 1K image.

What makes that pairing worth a second look is the per-unit transparency. Published per-second video pricing and fractional-cent image pricing lets a planner cost out an agent or batch workflow without guessing. The other angle is the conversational framing on Omni Flash, which Google describes as letting you refine and edit videos using natural language rather than a timeline UI. That, more than the 10-second clip length, is the part Google is leaning into.

Both models are live in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. Nano Banana 2 Lite is additionally rolling out today across Google consumer surfaces including AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads. The post is co-authored by Alisa Fortin and Anish Nangia, both Product Managers at Google DeepMind, who frame Nano Banana 2 Lite as their fastest, most cost-efficient Gemini Image model.

The honest caveat is that Omni Flash ships with real gaps that Google itself calls out. Audio references and scene extension are not yet supported, and video references up to 3 seconds are accepted by the API schema but not correctly processed by the model at this time. What the reporting does not give you is rate-limit guidance, token-level pricing for the editing prompts, or quality benchmarks against the prior Nano Banana model or rival systems. Those matter once you are sizing a real workload.

The thing to watch is whether the per-second video number holds once the missing pieces land. If it does, small teams and agencies suddenly have a short-form video model with predictable math behind it, and for most practitioners that is a more useful shift than another benchmark win.