Skip to content

Google cuts Gemini image-model pricing roughly in half

Nano Banana 2.1 replaces Nano Banana 2 as the flash-tier image model, adding 4K output and mask-based editing. The model it replaces is deprecated with no shutdown date yet.

Source: Google

Google shipped gemini-nano-banana-2.1 on 6 October, replacing Nano Banana 2 (gemini-3.1-flash-image) as its flash-tier image generation and editing model. It’s rolling out across the Gemini app, AI Mode in Search, AI Studio, Flow, Stitch, Google Ads and the Gemini Enterprise Platform.

The update adds mask-based editing for selecting and changing specific parts of an image, better subject consistency across multi-turn edits, sharper text rendering, output up to 4K, and new aspect ratios of 1:4, 4:1, 1:8 and 8:1. Multiple outlets covering Google’s own pricing page put the new per-image cost at roughly half of Nano Banana 2’s, from $0.0336 for a standard 1K image.

Why it matters: if you’re calling the Gemini image API, swapping the model ID to gemini-nano-banana-2.1 gets you cheaper output and more control over edits and aspect ratio without any other code changes.

The caveat: Google’s changelog marks gemini-3.1-flash-image deprecated but has not set a shutdown date, so there’s no forced migration deadline yet — but new projects should start on 2.1 rather than build against a model already marked for retirement.

Share

Related reading

ML for Web Devs

Tokens, context windows, and why your bill is what it is

The unit of billing for language models is not the request or the word. Once you understand tokens and how a context window fills up, the invoice stops being a surprise.

ML for Web Devs

RAG is just search with extra steps

Retrieval-augmented generation gets written up like an architecture. It is a search query, a string concatenation and one API call, and the search query is the part that decides whether it works.

ML for Web Devs

A prompt is a spec, so write it like one

Prompt engineering has very little to do with magic words. It is the same skill as writing a clear ticket for a capable contractor who will not ask you any questions.

More from The Wire

All briefs

Anthropic cuts distillation attacks on Claude by up to 99%

New safeguards, including a reasoning classifier and stricter account limits, now block most attempts to copy Claude by training on its output. Blocked requests in high-confidence cases are billed, and some integrations may need updating.

via Anthropic