Google has launched Nano Banana 2.1, a Gemini 3.6 Flash-based model for AI image generation and editing. The update lowers image-token API pricing from $60 to $30 per million tokens while adding improvements to composition, in-image text rendering and the ability to keep people and products visually consistent through successive edits.
The model can work with up to 14 reference images at once, preserving the appearance of up to four characters and 10 objects. It also supports mask-based editing, allowing changes to be limited to selected image areas rather than affecting an entire composition.
Google is rolling out the model across its consumer, advertising and developer products, including:
- Gemini app
- AI Mode in Google Search
- Google Ads
- AI Studio
- Flow
- Stitch
Nano Banana 2.1 supports 1K, 2K and 4K output, with Google also citing improved infographic generation and ultra-wide panoramas up to an 8:1 aspect ratio. The company says it has addressed artifacts that could appear in high-resolution panorama output. Developers can choose minimal, medium or high reasoning levels, and can connect generation to Google Search and Image Search for real-world information and imagery beyond user-uploaded references.
API image-generation pricing is listed at the following approximate rates:
- 1K image: $0.0336
- 2K image: $0.0504
- 4K image: $0.113
Google says the 1K and 2K rates are about half those of Nano Banana 2, while 4K generation is about 25% cheaper than the previous $0.151 rate. Batch processing further halves Nano Banana 2.1 pricing. These figures apply to image generation through the API; input processing and generated text are billed separately.
Google positions Nano Banana 2.1 as a more efficient alternative to Nano Banana Pro while retaining the speed associated with its Flash models. The rollout is gradual, so access may vary across the listed Google services.
