# Nano Banana 2 vs Nano Banana Pro vs Nano Banana 2 Lite: which one to use

By SeedRouter · Published 2026-09-25 · Updated 2026-09-25

Nano Banana 2, Nano Banana Pro and Nano Banana 2 Lite are Google's three current image models. Nano Banana 2 is Gemini 3.1 Flash Image, Nano Banana Pro is Gemini 3 Pro Image, and Nano Banana 2 Lite is Gemini 3.1 Flash Lite Image. All three take the same `generateContent` request body and return one image per request, so switching between them is a change of one `model` string.

The short answer: use **Nano Banana 2** by default, **Nano Banana Pro** when a complex prompt calls for Google's Pro-tier model, and **Nano Banana 2 Lite** for fast, high-volume 1K images.

## How do the three Nano Banana models differ?

|                        | Nano Banana 2                       | Nano Banana Pro                               | Nano Banana 2 Lite                   |
| ---------------------- | ----------------------------------- | --------------------------------------------- | ------------------------------------ |
| Google model           | Gemini 3.1 Flash Image              | Gemini 3 Pro Image                            | Gemini 3.1 Flash Lite Image          |
| Output sizes           | 512, 1K, 2K, 4K                     | 1K, 2K, 4K                                    | 1K                                   |
| Aspect ratios          | 14, including 1:4, 4:1, 1:8 and 8:1 | 10 standard ratios, 1:1 to 21:9               | 14, including 1:4, 4:1, 1:8 and 8:1  |
| References per request | Up to 14 images, videos or PDFs     | Up to 14 images                               | Up to 14 images, videos or PDFs      |
| Thinking control       | Model default                       | Thinks by default; thought summary on request | `thinkingLevel`: `MINIMAL` or `HIGH` |
| Input limit            | 131,072 tokens                      | 65,536 tokens                                 | 65,536 tokens                        |

Every figure in the table comes from the request validation each model runs on SeedRouter, and matches the [Nano Banana 2](https://seedrouter.ai/docs/gemini-3-1-flash-image), [Nano Banana Pro](https://seedrouter.ai/docs/gemini-3-pro-image) and [Nano Banana 2 Lite](https://seedrouter.ai/docs/gemini-3-1-flash-lite-image) API references.

## When should you pick Nano Banana 2?

Pick Nano Banana 2 when you are not sure. It has the widest range of output sizes, from a 512-pixel draft to a 4K final, and every aspect ratio the family supports, including the extreme 1:8 and 8:1 banners. It also reads videos and PDFs as references, so a product sheet or a short clip can guide the image.

Google describes it as the model designed for speed, efficiency and high throughput. In practice it is the one to build on first and move away from only when a job shows a reason.

## When is Nano Banana Pro worth it?

Nano Banana Pro is Google's Pro-tier image model and thinks through the prompt before it draws; you can ask for its thought summary with `includeThoughts`. That reasoning is what you want for prompts with many constraints, exact layouts or several references that must be combined.

Its limits are narrower. Nano Banana Pro takes image references only, not video or PDF, and it offers the ten standard aspect ratios, not the extreme banners. On the Official channel its token rates are the highest of the three, so compare the live prices on the [Nano Banana Pro page](https://seedrouter.ai/models/nano-banana-pro#pricing) before a large run.

## When does Nano Banana 2 Lite make sense?

Nano Banana 2 Lite is the family's low-latency model. It renders at 1K only, which is enough for thumbnails, feeds, previews and chat replies, and it keeps the full set of aspect ratios and reference types.

It is also the only one of the three that lets you set how much it reasons: `thinkingLevel` `MINIMAL` returns fastest, `HIGH` spends more time on a harder prompt. If your product shows the image while the user waits, start here.

## Does the choice change how you pay?

Nano Banana 2 and Nano Banana Pro are each sold on two channels. Standard IDs such as `gemini-3.1-flash-image` charge one flat price per delivered image. Official IDs such as `gemini-3.1-flash-image-official` bill the tokens each render reports, so larger sizes and longer thinking cost more. Nano Banana 2 Lite is sold on Official token billing only, as `gemini-3.1-flash-lite-image-official`. The channel decides the billing, the model decides the price level. The [Nano Banana API pricing guide](https://seedrouter.ai/blog/nano-banana-api-pricing) puts the live rates of all three side by side.

## What about the original Nano Banana?

The original Nano Banana is Gemini 2.5 Flash Image. Google's pricing page marks it as deprecated and says it will be shut down on October 2, 2026, and recommends moving to Gemini 3.1 Flash Image or Gemini 3.1 Flash Lite Image. New projects should start on one of the three models above.

## Frequently asked questions

### Is Nano Banana Pro better than Nano Banana 2?

It is Google's Pro-tier image model with thinking on by default, aimed at complex prompts. It is not better at everything: Nano Banana 2 has more output sizes, more aspect ratios and accepts video and PDF references. Test your own prompts on both in the [playground](https://seedrouter.ai/models/nano-banana-2#playground).

### Can I switch models without changing my code?

Yes. The request body is the same `generateContent` shape for all three. Change the `model` value and keep only options the new model supports, such as `imageSize` values and aspect ratios.

### Which Nano Banana model is fastest?

Nano Banana 2 Lite is the one Google designs for the lowest latency, and `thinkingLevel: "MINIMAL"` makes it faster still.

## Pick one and try it

Start with [Nano Banana 2](https://seedrouter.ai/models/nano-banana-2), move up to [Nano Banana Pro](https://seedrouter.ai/models/nano-banana-pro) for complex prompts, and use [Nano Banana 2 Lite](https://seedrouter.ai/models/nano-banana-2-lite) where speed matters most. One API key and one balance cover all three.