Google's Gemini Nano Banana 2.1 halves image output to $30

Nano Banana 2.1 succeeds Nano Banana 2 (Gemini 3.1 Flash Image), and Google says the new model improves image quality, prompt adherence, text rendering, and character consistency across multiple edits. It keeps Flash-class processing speed and cost efficiency while adding support for wide and panoramic images at 1:4, 4:1, 1:8, and 8:1 ratios in 2K and 4K, addressing the tiling artifacts that appeared in the earlier model.

The launch also rebalances how Google charges for the model. Image output drops to $30 per 1 million tokens from $60, but text and image input rises to $1.5 from $0.50, and text and thinking output rises to $7.50 from $3. For teams combining many reference images in a single request, the multi-image fusion capability now allows up to 14 reference images, maintaining consistency for up to four characters and fidelity for up to 10 objects, alongside grounding through Google Search and Google Image Search.

The practical question for adopters is how that mix lands on their bills, since cheaper image output may be offset by pricier input and text output depending on usage patterns. Google is also pressing users to migrate, as gemini-3.1-flash-image is deprecated and stops on October 29; whether teams move in time, and whether the quality gains justify the new pricing structure, is likely to shape how quickly Nano Banana 2.1 is adopted.