Nano Banana 2.1 Pricing: What Google's Image Model Costs

Nano Banana 2.1 is now generally available in the Gemini API, and the headline change is the price of a picture. Google's developer changelog lists the release on October 6, 2026, describing it as an update to Nano Banana 2 that keeps Flash-level speed while improving image quality. The pricing page shows what that means in practice: a standard 1K image now costs $0.0336, down from $0.067.

• Model ID: gemini-nano-banana-2.1, generally available since October 6, 2026.
• Per-image price: $0.0336 at 1K, $0.0504 at 2K and $0.113 at 4K on the standard tier.
• New formats: wide and panoramic aspect ratios of 1:4, 4:1, 1:8 and 8:1.
• Migration: Nano Banana 2 is marked deprecated, with no shutdown date announced.

What Is New in Nano Banana 2.1?

According to Google's changelog, the update improves visual quality, prompt adherence, character consistency across multi-turn edits, and text rendering. Those are the areas that matter most for conversational image editing, where a user asks for a change, then another, and expects the same character and layout to survive each round.

The more visible addition is format. Nano Banana 2.1 adds four wide and panoramic aspect ratios, 1:4, 4:1, 1:8 and 8:1, and supports output at 1K, 2K and 4K. An 8:1 frame is the shape of a website banner, a storyboard strip or a wraparound illustration, which previously meant generating several images and stitching them together.

For background on how this model family has developed, see our earlier coverage of Nano Banana 2 Lite for low-cost image generation and the June Gemini update that added thinking levels.

How Much Does Nano Banana 2.1 Cost per Image?

All figures below come from Google's Gemini API pricing page. The key change is that image output tokens are billed at $30.00 per million, down from $60.00 for Nano Banana 2.

• 1K image: $0.0336, down from $0.067, a cut of about 50 percent.
• 2K image: $0.0504, down from $0.101, also about 50 percent.
• 4K image: $0.113, down from $0.151, a cut of about 25 percent.
• Batch tier: $0.0168 at 1K, $0.0252 at 2K and $0.0567 at 4K.

So the halving applies to 1K and 2K output, while 4K sees a smaller reduction. At the standard rate, 1,000 images at 1K come to $33.60 instead of $67.

Two details are worth planning around. First, input pricing moves the other way: $1.50 per million input tokens, up from $0.50, and text output including thinking is $7.50 per million, up from $3. For workloads dominated by generated pixels the net result is still much cheaper, but pipelines that send long prompts or many reference images should rerun their own numbers. Second, the pricing page lists no free tier for the model, and the 0.5K size offered by Nano Banana 2 is not listed for 2.1.

Should Developers Migrate From Nano Banana 2?

Google's changelog marks the previous model, gemini-3.1-flash-image, as deprecated and recommends moving to the new ID. No shutdown date has been announced, so there is time to test rather than rush.

A sensible migration checklist is short. Regenerate a sample of your real prompts and compare character consistency across edits, since that is one of the stated improvements. Recalculate cost with your actual mix of input tokens and output resolution. And if you produce banners or strips, try the panoramic ratios directly instead of compositing.

The broader pattern is encouraging for anyone building with image models. Capability improvements are arriving together with lower per-image prices, which steadily widens the set of products where generated imagery is affordable at scale. I will be watching for independent quality evaluations, since the improvement claims so far are Google's own. More analysis of new model releases is in our AI coverage.

Sources: Google AI for Developers changelog — October 6, 2026; Google Gemini API pricing — accessed October 12, 2026; AI Weekly — October 7, 2026.