Google Launches Gemini 2.5 Flash Image for Smarter Editing

Google has unveiled its latest model for image generation and editing, known as Gemini 2.5 Flash Image or nano-banana. This update allows users to merge several images into one, maintain character consistency for storytelling, make specific changes using natural language, and utilise Gemini's extensive knowledge to generate and edit images.

The Gemini 2.5 Flash Image model is currently accessible through the Gemini API and Google AI Studio for developers, as well as Vertex AI for businesses. The pricing is set at $30.00 per million output tokens, with each image costing 1290 output tokens, translating to $0.039 per image.

Gemini 2.5 Flash Image Features

Significant updates have been made to Google AI Studio's "build mode" to facilitate easier use of Gemini 2.5 Flash Image, with more enhancements expected soon. Users can quickly test the model's capabilities using custom AI-powered apps, remix them, or bring ideas to life with a single prompt. Once an app is built, it can be deployed directly from Google AI Studio or saved on GitHub.

You can put the same character in different scenes, show a product from several angles, or make consistent brand visuals without altering the subject. The model lets you edit images with natural language. You can blur a background, clean up a shirt, cut someone out, change how a subject is posed, add color to old photos, and more - all with just one prompt.

Advanced Capabilities

Image generation models have focused on producing visually appealing results but struggled with real-world semantic understanding. Gemini 2.5 Flash Image changes that by leveraging Gemini's world knowledge, opening up entirely new applications.

The model can interpret and merge multiple input images, enabling users to insert objects into scenes, restyle rooms with different textures or color schemes, and seamlessly fuse images - all through a single prompt.

To broaden access, OpenRouter.ai has partnered with Google to deliver Gemini 2.5 Flash Image to over 3 million developers worldwide. Every image generated or edited with the model will also carry an invisible SynthID watermark, ensuring clear identification as AI-generated or modified.

Currently available in preview via the Gemini API and Google AI Studio, Gemini 2.5 Flash Image is expected to move to a stable release in the coming weeks.

Via