Overview

• Gemini 2.5 Flash Image is live globally in the Gemini chatbot on web and mobile with free image editing, and Adobe has begun trials in Firefly and Express.
• Developed by Google DeepMind, the model was internally known as “Nano Banana.”
• Users can carry out conversational, multiturn edits and combine up to three uploaded photos to create new scenes.
• The system prioritizes subject coherence to keep a person, pet, or product consistent across edits and angles, addressing earlier hallucination issues.
• Capabilities include background blur, object or person removal, pose changes, colorization, style transfer, and multi-image fusion, with outputs labeled by visible watermarks and an invisible SynthID marker.