[Digital Today reporter Jinju Hong (홍진주)] Google unveiled its image generation and editing artificial intelligence (AI) model, Nano Banana 2.1. It boosts performance while cutting developer image-generation costs to about half of the previous version, and expands its reach from the Gemini app to Search, Ads and developer tools.

On Oct. 6 (local time), major foreign media including blockchain outlet Decrypt reported that Google will roll out Nano Banana 2.1 sequentially across key services including the Gemini app, AI Mode in Google Search, Google Ads and Google AI Studio. Nano Banana 2.1 generates images from text or edits existing images, improving visual design quality, mask-based editing and subject consistency.

Google's own evaluation also confirmed performance gains. In a text-based image generation preference test, Nano Banana 2.1 scored 1,050 ELO points. Nano Banana 2 scored 990, while Nano Banana Pro scored 935. In an infographic factuality evaluation, Nano Banana 2.1 scored 0.521, well above Nano Banana 2's 0.179. At launch, results from external independent evaluations were not disclosed.

Its image editing range has also expanded. Nano Banana 2.1 can process up to 14 reference images at once and can track up to 4 people and 10 objects. It supports output resolution up to 4K and can also generate wide images with an 8 to 1 aspect ratio. It also strengthened its ability to preserve the original appearance of people or objects even after multiple rounds of editing.

It also added control functions for developers. Users can adjust the model's inference level from minimum to high before generating an image, and can set it to use Google Search and Google Image Search to verify information about people, objects and events included in images. It supports search-based factuality correction rather than stopping at simple image generation.

Prices were cut sharply from the previous version. Based on the developer API, the cost to generate a standard 1K image is $0.0336 per image, down by half from Nano Banana 2's $0.067. A 4K image costs $0.0756 per image, cheaper than the previous $0.151. An additional 50 percent discount applies to batch jobs. Generating 1,000 standard 1K images costs about $33.60.

Google will also deploy Nano Banana 2.1 in developer tools including AI Studio, Flow and Stitch. It is a strategy to target the enterprise and developer market alongside services for general users.

The Nano Banana series also affected the growth of Gemini. The first Nano Banana drew attention in September 2025 for a feature that turns selfies into collectible figure images, and Nano Banana 2 later introduced a function that checks search results first when depicting real people or events.

Nano Banana 2.1 adds performance improvements and price cuts on top of that. Attention is focused on how many developers and enterprise customers it will draw to the image-generation AI market by emphasizing search integration, 4K output, multiple reference images and lower API costs.

Meet Nano Banana 2.1, our latest image generation and editing model This upgraded version outperforms our previous models across the board, with notable leaps in visual design, mask-based editing, and subject consistency to help you create more natural-looking images.