Nano Banana: Google powers Gemini app with enhanced image generation model

Called Gemini 2.5 Flash Image, the Nano Banana model debuts with features like character consistency, multi-turn editing, and image blending, available via API, AI Studio, Vertex AI, and Gemini app

Google has announced Gemini 2.5 Flash Image, its latest image generation and editing model, alongside a major update that brings these capabilities directly into the Gemini app. According to the US-based software giant, the new model, dubbed Nano Banana, is designed to maintain consistency across edits and support more advanced editing tools.

Gemini 2.5 Flash What’s new

The Gemini 2.5 Flash Image model introduces several new capabilities:

• Character consistency: Ensures that people, pets, or objects look the same across multiple images or edits, useful for storytelling, product visuals, or brand assets.
• Prompt-based editing: Allows natural language commands for specific edits such as changing backgrounds, removing objects, or altering poses.
• Native world knowledge: Uses Gemini’s broader understanding to interpret diagrams, answer questions, or follow complex editing instructions.
• Multi-image fusion: Supports combining multiple input images, letting users place objects into scenes or restyle environments with new textures and colors.

Nano Banana image editing model is now in Gemini app

Google has expanded these image editing features to the Gemini app. Users globally, both free and paid, can now:

• Maintain likeness across edits: Keep the same look for people, pets, or objects while applying changes like new outfits or hairstyles.
• Blend multiple photos: Merge different photos into a single scene, such as combining portraits of people and pets.
• Apply multi-turn edits: Make step-by-step changes to an image, including adding furniture to a room or altering specific parts while preserving others.
• Mix designs and styles: Transfer the colour, pattern, or texture from one image onto objects in another.

Notably, all images edited or generated in the Gemini app include both a visible watermark and an invisible SynthID watermark to indicate they were created with AI.

For developers and enterprises

Developers and enterprises can access Gemini 2.5 Flash Image through Google AI Studio.

Pricing

Gemini 2.5 Flash Image is available via the Gemini API, Google AI Studio, and Vertex AI. The model is priced at $30 per 1 million output tokens, with each generated image consuming 1,290 output tokens, equivalent to about $0.039 per image. Other input and output modalities follow standard Gemini 2.5 Flash pricing, said Google.