Google has just released Gemini 2.5 Flash Image, a new image editing and generation tool woven into the Gemini app. Known as the cryptic “nano banana” during testing, this model has already impressed early users, topping leaderboards and setting the bar high for what AI can do with visuals.
For months, whispers of “nano-banana” have been circulating online and on testing platforms like LMArena, where it appeared anonymously and wowed users. Google has now confirmed those rumors, and the mystery model is indeed Gemini 2.5 Flash Image, built by the DeepMind team. Unlike earlier AI image editors that often struggled to keep a subject’s identity intact across changes, this model is all about consistency. Upload a photo of your friend, your dog or yourself and you can put them in a 1960s diner or a futuristic cityscape and the subject will still be themselves.
No products found.
Editing with Gemini 2.5 Flash Image is as simple as typing a sentence. Need to swap out a plain T-shirt for a Hawaiian one? Easy. The model understands natural language prompts so you don’t need to be a photo editing expert to get pro results. It can even add color to black and white images or change a person’s pose with a few words.
Beyond single image adjustments, the model is great at combining many images into one scene. Upload a photo of yourself and another of your pet and then ask Gemini to make a portrait of you both stargazing on a mountain. The results are surprisingly realistic with the model retaining face features and fur patterns throughout the blend.
Multi-turn editing is another cool feature which allows you to edit an image step by step. Start with an empty room, then paint the walls blue, add a bookshelf and finally a armchair – all with simple text instructions. Each change builds on the previous one while keeping the overall scene intact. This feels like working with an artist who understands your idea and keeps it no matter how many changes you make.
Google’s model doesn’t just stop at editing existing photos; it can apply styles to images. Use the texture of flower petals to rebuild a pair of rainboots, or a butterfly’s wing pattern to create a garment. This feels like a magic wand for creation, allowing you to play with aesthetics in ways that were previously exclusive to designers.
The Gemini 2.5 Flash Image model is available to both free and paid users of the Gemini app, with free users able to create up to 100 edits daily and paid users getting ten times that. Developers can access it through the Gemini API, Google AI Studio and Vertex AI, for $0.039 per image generated. Adobe has already jumped on board, integrating the model into Firefly and Express, with full access for Express users starting September 1.
Every image created or edited with this model has a visible “AI” watermark and an invisible SynthID digital watermark so you know it’s AI. This is about responsible use, addressing the concerns around misuse in an era where realistic fakes are a growing problem.
