Google today announced Gemini 2.5 Flash Image (tested under the name "nano-banana"), its state-of-the-art image generation and editing model. Google claims that this new model can generate and edit images while maintaining character consistency and blending multiple images into a single one.
According to LMArena, the Gemini 2.5 Flash Image is now the top-rated image editing model in the world, beating OpenAI"s GPT Image 1 and Flux.1 Kontext. In general, previous image generation models performed better aesthetically but lacked an accurate semantic understanding of the real world. The new Gemini 2.5 Flash Image takes advantage of Gemini’s world knowledge to create more accurate and realistic images.
This new Gemini 2.5 Flash model is available for both consumers and developers. For developers, this model is available via the Gemini API, Google AI Studio, and Vertex AI for enterprise. For developers, Gemini 2.5 Flash Image is priced at $30.00 per 1 million output tokens. So, it would cost around $0.039 per image.
For consumers, this model can be accessed via the Gemini web and mobile apps. Google highlighted the following use cases for general consumers to try on the Gemini app:
- Give yourself a costume or location change: Upload a photo of a person or pet, and the model will keep their look the same in every image as you place them in new scenarios.
- Blend photos together: You can now upload multiple photos and blend them together for a brand-new scene. For example, take your photo and another of your dog to create a perfect portrait of you both on the basketball court.
- Try multi-turn editing: You can keep editing the images Gemini makes — take an empty room, paint the walls, then add a bookshelf, some furniture, or a coffee table.
- Mix up designs: Apply the style of one image to an object in another. You can take the color and texture of flower petals and apply it to a pair of rainboots, or design a dress using the pattern from a butterfly"s wings.
Google claims that it is not using the actual images you upload to train its generative machine-learning technologies unless they"re included in feedback. All images created or edited in the Gemini app will have a visible watermark and an invisible SynthID digital watermark.