Google Introduces Nano Banana 2.1 Model with Enhancements

On October 6, 2026, Google released a new version of its artificial intelligence model, Nano Banana 2.1, which is responsible for generating images based on text prompts and editing existing photos. The updated version is being integrated into the Gemini app, Google Search AI mode, Google Ads, and development tools such as Google AI Studio, Flow, and Stitch. According to the company, the updated Nano Banana 2.1 model surpasses previous models in all areas, including improvements in visual design, mask-based editing, and object sequencing. “This updated version surpasses our previous models in all areas, with noticeable improvements in visual design, mask-based editing, and object sequencing to help you create more natural images,” Google representatives noted. In Google's own tests, the model showed an infographic factual accuracy score of 0.521 compared to 0.179 for the previous version. Independent testing at the time of launch is unavailable.

The first Nano Banana made Gemini a hit. In September 2025, its trick of turning selfies into collectible figurines propelled the app to the top spot in both major stores, displacing ChatGPT after nearly three years of leadership. Alphabet's market value exceeded $3 trillion during the same period. Version 2 was released in February based on Gemini 3.1 Flash Image. Its main feature was the verification of Google Search before rendering real objects, increasing the likelihood of accurately depicting a real event or person.

In tests for text-to-image conversion, Nano Banana 2.1 scored 1,050 ELO points, higher than Nano Banana 2 with 990 points and Nano Banana Pro with 935 points. The new version can process up to 14 reference images simultaneously and track up to four characters and ten objects. It can output images with a resolution of up to 4K, comparable to television quality. Developers can also adjust how long the model "thinks" before rendering, from minimal to high levels, and allow it to rely on Google Search and Google Image Search.

The cost of using the API for developers is $0.0336 for a standard 1K image, which is half the price of Nano Banana 2 at $0.067. For a 4K image, the price is $0.0756 compared to $0.151 for a similar image on Nano Banana 2. Batch jobs, or large volumes of images processed in batches, receive an additional 50% discount. As a result, 1,000 images now cost about $33.60 through the API, significantly cheaper than $67 on Nano Banana 2, making this model very efficient and cost-effective.