Google LLC said today it’s updating its Gemini app and chatbot with a powerful new artificial intelligence image model that will give users fine-grained photo editing capabilities.
The new model, named Gemini 2.5 Flash Image, debuts today on the Gemini app so that users can edit their photos with natural language. The company claims the new model provides state-of-the-art image generation and editing for photos, keeping the original composition, objects and people intact while adding, changing or removing whatever the user wants.
Google DeepMind, the company’s AI research arm, tested the new model under the mysteriously silly name “Nano Banana” on LMArena. This public site crowdsources anonymous feedback on AI model quality. At first, it was unknown what this new model might be, especially given its weird name, but users quickly sussed out that it must be from Google.
The model outperformed every other photo-editing model on the site in early preview, gaining it the title of “top-rated editing model in the world.” Although it was not without its flaws, the model proved to be superior for consistency, quality and following instructions.
Now, DeepMind has revealed that the model is actually Gemini 2.5 Flash Image and it powers the new Gemini image editing experience.
“We’re really pushing visual quality forward, as well as the model’s ability to follow instructions,” Nicole Brichtova, a product lead on visual generation models at Google DeepMind, said in an interview with News.
One of the problems for image editing and AI models has always been that models tend to make subtle or large modifications to images, even when users ask them to make small changes. For example, users might take a photograph of themselves and ask a model to add glasses. The model might add glasses to their face, but it could dramatically change their features, adjust their hairstyle or an object in the background might change from one thing to another.
To test out the new model, Google suggested people try it out with a photo of themselves. They could use it to put themselves in a new outfit or change their location. The model is also capable of blending subjects from two different photos into a brand-new scene — for example, taking a picture of you and your cat and having it put you together on the couch.
According to Google, this new model allows users to make multi-turn edits: Just take a photo and ask for one change and then follow up with another. This allows for iterative modifications to photos or images that feel natural. Since prompts can make specific requests about locations or subjects, the model will only change them and nothing else.
Developers can also get access to this capability through Google’s AI platforms and tools via Gemini API, Google AI Studio and Vertex AI.
In addition to the news of the release, Adobe Inc. announced that it will add the new model to its AI-powered Firefly app and Adobe Express, making it easier for users to use the model to modify their photos and create stylized graphics with a consistent look and feel.
Images: Google
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
- 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more
- 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network.
About News Media
Founded by tech visionaries John Furrier and Dave Vellante, News Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.
