
Google has introduced Gemini 2.5 Flash Image, a new image generation and editing model designed for quick, high‑quality creativity across personal, developer, and enterprise use cases. The model, informally called “nano‑banana,” focuses on keeping characters and styles consistent across multiple images, following visual templates, and making precise, local edits using simple natural‑language instructions. It is aimed at creators who want fast turnarounds with better control over details like texture, color, pose, and composition.
According to Google, the model supports targeted transformations and multi‑turn editing, so users can continue refining an image through a back‑and‑forth conversation. Typical examples include changing outfits or locations while preserving the subject’s identity, blending several photos into a new scene, and applying the style of one image to objects in another such as using a flower’s color and texture to redesign boots or garments. The company adds that the model’s improved world understanding helps it respect scene logic and deliver consistent results across a series of prompts.
For everyday users, the model is available inside the Gemini app, making it easy to create or edit pictures with conversational prompts. Google highlights practical trials like costume or background swaps for people and pets, multi‑image fusion for quick scene design, and iterative edits that refine lighting, background cleanup, or object additions. These capabilities are intended to reduce manual steps and speed up content production for social posts, product shots, and personal projects.
For developers and businesses, Gemini 2.5 Flash Image can be accessed through the Gemini API, Google AI Studio, and Vertex AI, enabling teams to prototype, integrate, and deploy visual workflows at scale. Google lists pricing at USD 30.00 per one million output tokens; each generated image uses 1,290 output tokens, which comes to about USD 0.039 per image. All content created or edited in the Gemini app includes a visible watermark and an additional invisible SynthID digital watermark to clearly indicate AI‑generated or AI‑edited media.
With its focus on character consistency, fine‑grained edits, and conversational control, Gemini 2.5 Flash Image steps forward as a fast, affordable option for creators, developers, and brands looking to build cohesive visual stories at very flash speed.
Source: Google Blog