Google’s DeepMind team pushed a new image model into general availability on June 30, and the internal-nickname-turned-public-brand really is Nano Banana 2 Lite. The technical model ID is Gemini 3.1 Flash-Lite Image. The pricing is $0.034 per 1,000 images. The latency target is roughly four seconds. It is available immediately in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. Alongside it, Gemini Omni Flash entered public preview as a natively multimodal video generation and editing model that aims to slot into the same enterprise agent workflows.
The name is a joke that got promoted, and Google is fully committed to it now. The rest is not a joke. Three cents per thousand images at four-second latency is not a hero-shot model, which is exactly the point. This is the tier meant for retail catalog refreshes, real estate photo variants, ad creative A/B batches, and every internal tool that needs a picture generated in a loop somewhere and does not care about winning any awards. Every hallway conversation about “we should just have an AI generate that” now has a supported line item under it.
The strategic read is that image generation is finishing its move from headline feature to commodity utility, on roughly the same trajectory that transcription took after Whisper. The public splashy image announcements will keep going for a while at the frontier tier, where a Nano Banana 2 non-Lite still lives at higher prices. Underneath, the boring high-volume use cases are being priced flat enough that no procurement conversation is going to slow them down. Omni Flash in public preview is the same story starting for video. That path takes longer, but the arrow points the same direction.
Somewhere in the DeepMind org chart, someone tried to un-name it. They lost.