Google on Tuesday introduced Gemini 2.5 Flash Image, an artificial-intelligence model for image generation and photo editing that is now embedded in the consumer Gemini app and available to developers through the Gemini API, Google AI Studio and Vertex AI. The system, which Google tested under the code name “Nano Banana,” is billed as the company’s most capable visual model to date. Developed by Google DeepMind, the model leads the public LMArena leaderboard for image editing and is designed to keep a subject’s likeness intact while applying complex transformations. Users can blend multiple photos, apply the style of one image to objects in another, and make multi-stage edits using plain language prompts. All content carries visible and invisible SynthID watermarks to indicate AI provenance. Gemini 2.5 Flash Image is free for all Gemini consumers, while cloud access is priced at $30 per million output tokens—about 3.9 cents per image—according to Google’s published rates. Product lead Nicole Brichtova said the release “pushes visual quality forward” and reflects months of feedback gathered during the model’s anonymous public testing. The launch intensifies competition with established creative-software providers. Adobe said it will incorporate the model into its Firefly and Express tools, and industry observers noted that Gemini’s ability to perform tasks once reserved for applications such as Photoshop raises the bar for rival platforms including OpenAI’s DALL-E, Midjourney and Ideogram.