Google Launches Nano Banana 2.1: Better Than Its Predecessor at Half the Price

Google introduced Nano Banana 2.1, an incremental update to its text-to-image and image-editing model, and has started deploying it across consumer and developer surfaces including the Gemini app, Search’s AI Mode, Google Ads and tools such as Google AI Studio, Flow and Stitch. The company highlighted three main improvements: enhanced visual design, more precise mask-based editing (so edits are limited to marked regions), and greater subject consistency across multiple edits. According to Google’s internal tests, Nano Banana 2.1 outperformed prior versions on several measures. In ELO-style preference tests, the new model scored 1,050 points versus 990 for Nano Banana 2 and 935 for Nano Banana Pro. Google also presented an infographic factuality score of 0.521 for 2.1 compared with 0.179 for earlier variants. The company notes the model can handle up to 14 reference images at once, track up to four characters and ten objects, and produce outputs up to 4K resolution and as wide as an 8:1 aspect ratio. Developers can tune how long the model “thinks” before rendering and can allow it to ground outputs by consulting Google Search and Google Image Search. Google priced the model’s API access substantially lower than before: a standard 1K image is $0.0336 (about half the $0.067 charged for Nano Banana 2), and a 4K image is $0.0756 compared with $0.151. Batch processing receives an additional 50% discount. Google’s performance claims are based on the company’s own evaluations; the company did not provide independent third-party test results at launch. The update follows the original Nano Banana’s role in boosting Gemini’s popularity in 2025 and the February release of Nano Banana 2, which added web-grounding to improve accuracy when depicting real-world subjects.