Google unveils Nano Banana 2.1 image generation model

Summary

Google has released Nano Banana 2.1, built on the Gemini 3.6 Image infrastructure. This updated AI image generation and editing model focuses on improved visual composition, mask-based local editing, and character/object consistency across images. The model is available on Gemini, Google Search AI Mode, AI Studio, Flow, Stitch, Google Ads, and Gemini Enterprise platforms. Developers can access it with resolutions ranging from 1K to 4K.

Why it matters

In generative image tools, maintaining character and product consistency across multiple generations is one of the most requested yet least solved problems. The specific focus on this area could transform workflows in advertising, e-commerce, and content production. The fact that the model is available simultaneously across all of Google's product surfaces (search, ads, enterprise) indicates an intention to make it a default layer rather than a standalone tool. This raises the expectation bar for competitors like Midjourney and DALL-E.

Highlights

• Mask-based editing allows users to modify specific regions without regenerating the entire scene, reducing the burden of prompt rewriting.
• Consistency of character or product identity across consecutive generations is explicitly targeted, addressing a long-standing issue.
• The developer API (gemini-nano-banana-2.1) accepts text, image, audio, and video inputs and produces outputs up to 4K resolution.

Critical take

The article is essentially a near-verbatim repetition of Google's announcement. It lacks independent benchmarks, side-by-side comparisons, or failure scenario tests. Consequently, the claim of 'significant improvements' remains entirely vendor-sourced. The transition from Gemini 3.1 Flash Image to Gemini 3.6 Image is mentioned without any architectural explanation, preventing readers from assessing whether the gains are incremental or structural.