If you want to tweak the composition or a person's look after generating an image, do you have to regenerate from scratch every time? Google's Nano Banana 2.1 supports image generation and editing in the Gemini API, and it is a model that can use multiple reference images and search grounding together.
The model ID is gemini-nano-banana-2.1, and the Google Cloud model docs list general availability (GA) and the release date as October 6, 2026. Below, we look at the image generation and editing features and the API pricing.
Source: X @Google official announcement
Source: X @Google product rollout notice
A follow-up reply to Google's official announcement listed the Gemini app, Search AI Mode, AI Studio, Flow, Stitch, Ads, and Gemini Enterprise as the products it applies to. That is the list of rollout targets Google's announcement gave. How it is offered and the detailed conditions may differ by product, and the model going GA on Google Cloud Agent Platform and the rollout to the Gemini Enterprise app are separate items.
What kind of model is Nano Banana 2.1?
Nano Banana 2.1 is an image generation and editing model available in the Gemini API. You can make a new image from a text prompt or edit an image by providing it as input, and it also supports refining results over the course of a conversation.
Google says it improved visual quality, prompt adherence, character consistency across multiple turns, and text rendering inside images compared with the previous model. It is especially useful for work that has to keep the same person or character across several scenes and for making materials where the wording in the image matters.
Up to 14 reference images and search grounding
Depending on the request, you can use input images and reference images together, and according to the docs, reference images go up to 14. You can use this to follow the style of a product photo or to build a new scene from images of a person from several angles. It supports both generating images and editing existing ones.
You can also use Google Web Search and Image Search grounding. Context based on search results can be reflected in the generation process, so you can connect it to work that draws on current information or visual references. In actual use, you have to specify the tools and output resolution you need in the API request.
Nano Banana 2.1 API pricing
Gemini API Standard rates are split by input and output type. Input costs $1.50 per million tokens for text, image, and video tokens, text and thinking output costs $7.50 per million tokens, and image output costs $30 per million tokens. The official docs support 1K, 2K, and 4K output resolutions, and the table below lists the cost per image at 1K and 2K.
| Output resolution | Standard, per image | Batch, per image |
|---|---|---|
| 1K | $0.0336 | $0.0168 |
| 2K | $0.0504 | $0.0252 |
The Batch image output rate is $15 per million tokens, half of Standard. The per-image cost is also half at each resolution. If you have a lot of work and a flow that suits batch processing, it is worth comparing costs.
Comparing image output rates across models, Nano Banana 2.1 is half of Nano Banana 2 Standard and a quarter of Pro. This comparison is based on the image output token price, and the actual cost of a request varies with input tokens, output resolution, and usage.
What can you use it for?
- Making new images: Create image drafts for content from a text prompt.
- Editing existing images: Input the original and specify the element you want to change in conversation.
- Composing with references: Pass people, products, and style references through up to 14 images.
- Search-based generation: Connect Web or Image Search grounding to a request.
Pricing depends on the processing mode, Standard or Batch, and the image resolution. If you first decide whether you really need a high-resolution result and whether you can process several images in a batch, it is easier to estimate costs.
Read it alongside the official docs
You can check the model's input and output features and settings in the Gemini API Nano Banana 2.1 model docs. For per-token and per-resolution rates, see the Gemini API pricing page. Google Cloud's model detail page gives the GA status and release date.
To look at changes to Gemini app features alongside this, see the Gemini Skills launch news, and for updates to Google Search features, check Google AI Mode's web information tracking alerts.