Google Launches Nano Banana 2.1 With Better Image Editing and Lower Costs

Google has unveiled Nano Banana 2.1, an updated image generation and editing model powered by Gemini 3.6 Flash, at a lower cost and offering improved consistency. Released on Tuesday, the model expands across Google's products like Gemini app, AI Studio, API, Search AI Mode, Ads, Flow, and Stitch. Nano Banana 2.1 enables text-to-image generation and editing with up to 1 million tokens of context and supports 1K, 2K, and 4K resolution outputs.

The model excels in visual quality, text rendering, mask-based editing, and maintains consistency during repeated edits involving the same subjects. A key feature is multi-image editing, where it can process up to 14 reference images while keeping consistency for four characters and fidelity for ten objects. For developers, Google has slashed image-output pricing by half, now costing $30 per million image tokens on the standard paid Gemini API tier.

This price reduction, coupled with Google's higher Elo scores in text-to-image preference and editing evaluations, suggests Nano Banana 2.1 could become a routine component in software products. However, the model is not without limitations, including potential hallucinations, occasional timeouts, and weaknesses in small-text rendering and world knowledge.

Enterprise teams and API builders must migrate endpoints and benchmark output quality within a three-week window, as Nano Banana 2 will be deprecated by late October.