Jeremy Bowen
Now, users can enjoy greater world knowledge, quality, and thinking power in Nano Banana.
Google has released Nano Banana 2, the newest version of its AI image creation model. The company says this new tool is both faster and better than the earlier version. It combines the Pro series’ smart capabilities with the speed of Gemini Flash. Nano Banana 2 includes improvements like quicker image generation, clearer text display, better subject consistency, and ready-to-use results. It is also available in more countries and languages.
Google has launched Nano Banana 2 with the promise of faster performance and higher-quality visuals. The update also improves text handling and consistency, and it is now available across several Google products and services.
The original Nano Banana, officially known as Gemini 2.5 Flash Image, became widely used in the Gemini platform. Nano Banana Pro, also called Gemini 3 Pro Image, introduced more accurate text display and stronger real-world understanding last year.
Nano Banana 2 is the next step in Google’s image generation tools, building on Nano Banana and Nano Banana Pro.
Intelligence and visual quality at Flash speed
Nano Banana 2 brings the fast intelligence of Gemini Flash to image creation. It allows quick edits and repeated improvements.
- Advanced world knowledge: The model uses Gemini’s real-world knowledge and combines it with up-to-date information and images from web searches to create more accurate visuals. This deeper understanding helps users design infographics, turn notes into diagrams, and create data visuals.
- Precision text rendering and translation: Nano Banana 2 can generate clear and readable text for marketing designs or greeting cards. It can also translate and adapt text inside an image so ideas can be shared globally.
The main improvement in Nano Banana 2 is the balance between speed and quality. It creates and edits images quickly while still producing clear details and rich textures.
Real-world knowledge means using live information from web searches to make images more accurate and relevant.
Another key upgrade is text rendering, which allows the model to place clear and correct words directly into images. This is useful for marketing designs, posters, and diagrams.
Nano Banana 2 also supports in-image localisation, meaning text inside a picture can be translated or adjusted for different languages without losing clarity.
Enhanced creative control
Nano Banana 2 also greatly reduces the gap between speed and visual quality. It produces high-quality, realistic images.
- Subject consistency: It can keep the look of up to five characters and maintain up to 14 objects in one workflow. This helps users create storyboards and visual stories without changing the appearance of their inputs.
- Precise instruction following: With better instruction handling, the model follows detailed requests more closely, capturing the exact idea so the final image matches what was asked.
- Production-ready specs: Users can create eye-catching content with control over aspect ratios and resolutions from 512px to 4K, ensuring visuals stay sharp for both social media posts and large displays.
- Visual fidelity upgrade: Nano Banana 2 delivers brighter lighting, richer textures, and sharper details while maintaining the fast performance expected from Flash.
Google has also highlighted subject consistency as an important feature. It allows the model to keep the identity of characters or objects the same across multiple edits. For example, if a user creates a series of images with the same person, the model works to keep their appearance consistent.
Try Nano Banana 2 today
Whatever your needs, there is now a suitable tool for every workflow: Nano Banana Pro for high-quality tasks that need maximum accuracy, or Nano Banana 2 for fast creation, precise instruction following, and built-in image-search support.
Nano Banana 2 is being added across many Google products. It is now the default image model in the Gemini app for creating and editing images. It is a powerful AI Mode in Google Search and image features in Google Lens.
Nano Banana 2 will be available in preview through the Gemini CLI, Gemini API, and Vertex API for developers. It will also be accessible through AI Studio and the company’s development tool Antigravity.
Robust provenance: marking and verification
As AI-generated media grows, the tools used to identify it must also improve. Google continues to strengthen its tracking approach by combining its SynthID technology with C2PA Content Credentials, giving users clearer insight into whether AI was used and how it was used.
Along with performance improvements, Google continues to focus on content tracking. Images created by the model include an invisible digital watermark using the SynthID system. This helps indicate that the image was produced by AI and allows it to be verified later using checking tools.
Since its launch in November, the SynthID verification feature in the Gemini app has been used more than 20 million times across different languages, helping people identify AI-generated images, video, and audio.
