Google ships Nano Banana 2.1 image model at a lower price point

Google refreshed its popular image model. Logan Kilpatrick, who leads AI Studio, announced Nano Banana 2.1 as 'a nice update for image generation and editing customers'. He highlighted higher-quality images, many bug fixes from previous versions, and a new lower price point. It is available immediately in the Gemini API, AI Studio and the Gemini app, and DeepMind published a technical model card alongside it.

The Nano Banana line has become Google's consumer-facing image engine, known for consistent character editing and fast iteration. A point release that cuts price while improving quality targets developers building high-volume image pipelines, such as e-commerce product shots, marketing variants and app features, where cost per image decides which model gets picked.

The competition is intense. OpenAI is simultaneously turning ChatGPT's image-generation wait time into ad inventory, and open video and image models such as Kandinsky 6.0 keep improving under permissive licenses. Lower pricing helps Google defend its API share, and the model card adds the transparency that enterprise buyers increasingly expect.

This release is separate from Google's restricted rollout of Gemini 4 Argon to vetted cybersecurity researchers and government testers, which reflects a different strategy for frontier capabilities. Watch for third-party image-arena rankings, published per-image pricing comparisons, and whether the quality gains show up in editing consistency, the model's main selling point.