Build with Nano Banana 2, our best image generation and editing model

Today, we’re introducing Nano Banana 2 (Gemini 3.1 Flash Image), which brings high-fidelity image generation and faster, advanced editing to the new Flash model. It allows you to deploy sophisticated visual creation at scale with an amazing price-performance ratio.

Here’s a look at the new capabilities you can build with in Google AI Studio or the Gemini API. A paid api key is required to use the model on Google AI Studio.

Improved world knowledge

Nano Banana 2 leverages the Gemini model’s broad world knowledge to create enhanced visuals using images via web search. This allows developers to generate more detailed depictions inspired by real-life references.

To show this visual grounding in action, we built an example app called “Window Seat” that uses Nano Banana 2’s knowledge and web image search to create photorealistic window views inspired by world locations and live weather data.

Advanced text rendering and localization

Building dynamic UI generators or creative tools requires text that is crisp and accurate. Nano Banana 2 delivers an upgrade from previous Flash image models with more reliable text rendering.

Whether you're building applications that produce images at scale or creative apps, text can now render with the same precision as the artwork. Nano Banana 2 also supports in-image localization, allowing you to generate or translate text across multiple languages directly within the image.

To show these capabilities, we built a demo app “Global Ad Localizer” that translates an advertisement into different languages for international markets. Not only does this highlight the model's advanced in-image localization by rendering translated text, it simultaneously understands and localizes the visuals as well.

Greater creative control and consistency

Nano Banana 2 delivers speed and visual fidelity, with vibrant lighting, richer textures and sharper details. You now have more control over these high-fidelity outputs with several advanced creative features:

• Native aspect ratios: Seamlessly generate or edit images to fit your project requirement, with native support for all existing aspect ratios we have now added 4:1, 1:4, 8:1 and 1:8. Find the full list of aspect ratios supported.
• New 512px resolution: Optimize for efficiency. Joining our 1K, 2K and 4K offerings, this new resolution tier minimizes latency for rapid-fire iterations and heavy-duty pipelines.
• Improved instruction following: The model adheres much more strictly to complex, multi-layered developer prompts, helping to ensure the image your application requests is what is generated.
• Configurable thinking levels: You now have more control over the model's reasoning. Adjust thinking levels (Minimal (default) vs. High/Dynamic) to allow the model to reason through complex prompts before rendering, significantly improving output quality and prompt adherence.

Our “Pet Passport” demo takes a single photo of your pet and sends them on a global adventure to famous landmarks. It maintains your pet's appearance across different global destinations. For a truly custom output, we have included different creative controls settings.

Built for production: what developers are saying

We are already seeing incredible innovation from partners who have integrated the model into their production workflows to scale their visual features:

Build with Nano Banana 2 today

You can start building via a paid api key with Nano Banana 2 today. The model is available now via the Gemini API in Google AI Studio, for enterprise deployment on Vertex AI. It is available for use in Google Antigravity and Firebase. Here are a few places to get started:

• Jump right into the Gemini API developer docs
• Explore the app gallery and build apps in Google AI Studio with Nano Banana 2
• Experiment and test out the model in Google AI Studio
• Dive into the cookbook

We’re excited to see what you build with Nano Banana 2.