Google priced image and video generation by the unit — read the price list

On June 30, 2026 Google shipped two media-generation models. Nano Banana 2 Lite (gemini-3.1-flash-lite-image) does text-to-image in about 4 seconds at $0.034 per 1,000 images — the fastest in the Nano Banana family. Gemini Omni Flash (gemini-omni-flash-preview) generates and edits video conversationally at $0.10 per second of output, currently capped at 10-second clips (no audio references, no scene extension, no video references over 3 seconds yet). Both run in AI Studio, the Gemini API, and the Enterprise Agent Platform; Nano Banana 2 Lite is also rolling into Search, the Gemini app, Photos, and Ads.

Photo by Jakub Żerdzicki on Unsplash.

What I take from it

I read launches like this as a price list, not a demo reel. The capability is assumed now; the real question is what the unit cost makes affordable to build.

• $0.034 per 1,000 images changes the default. At three-and-a-half cents per thousand, generated imagery stops being a feature you ration and becomes something you can put in a loop — thumbnails, variations, per-record art. The constraint moves from cost to taste.
• $0.10/second of video is a budget you can actually reason about. A 10-second clip is a dollar. That's cheap enough to A/B, expensive enough that you don't generate blindly. I'd meter it like any other per-call cost and cache aggressively.
• The limits are the spec. "No scene extension, 10-second cap, no audio references" isn't fine print — it's the boundary of what you can promise a user today. I design to the documented limit, not the demo.

The pattern I keep seeing in 2026: the interesting model news is in the pricing table. Capabilities converge; unit economics decide what actually ships. A four-second image at three cents a thousand is a different product decision than the same image at ten times the price.

Sources: Start building with Nano Banana 2 Lite and Gemini Omni Flash (Google).