Google's Gemini 2.5 Flash Image model is now available for production use. The model can generate, edit, and combine images.
Gemini 2.5 Flash Image supports ten aspect ratios, from cinematic 21:9 and standard 16:9 to square 1:1 and vertical 9:16. Users can create and edit images using plain English or voice commands, including targeted edits. Images can be exported without captions or extra text.
Pricing starts at $0.039 per image, and one million output tokens cost $30. Additional pricing matches the standard Gemini 2.5 Flash model. The model is available through the Gemini API and Vertex AI.
Developers can build and test apps in Google AI Studio. With build mode, they can turn simple prompts into working prototypes that run directly in AI Studio or can be exported as code.
Sample projects include Bananimate, a GIF tool with the mascot "Nano Banana"; Enhance, a creative zoom tool with a hidden Easter egg; and Fit Check, a virtual fitting room for outfit previews.
Real-world use cases
The model is a good fit for projects that need consistent character design and flexible image processing. Startup Cartwheel combines Gemini 2.5 Flash Image with its 3D posing tool, so users can render characters from any angle. Co-founder Andrew Carr says other models struggle with either perspective or context, but Gemini 2.5 Flash Image handles both at the same time.
Volley, an AI studio, uses the model in its game "Wit's End." The game generates portraits, scene transitions, and image edits on demand. CTO James Wilsterman says latency is under ten seconds, so players can control everything in real time using voice or chat.