Google has released its newest artificial intelligence (AI) model, ‘Gemini 2.5 Flash Image’, which is built to generate and edit images with greater accuracy and flexibility. This latest model introduces advanced editing features like natural language prompts, the ability to merge multiple photos into a single composition, and improved consistency when working with faces, objects, and complex scenes.
Notably, earlier AI tools often had trouble keeping faces accurate or holding on to small details when images were changed, but Gemini 2.5 Flash Image (aka nano-banana) is built to manage those tasks more reliably. Users can give conversational prompts (in natural language) and the system produces seamless results that keep people, pets, or products recognizable.
The model uses the wider Gemini system’s knowledge to make edits that are not only realistic but also accurate in context. Even the model can apply cultural, historical, or real-world awareness to the creative process. Another prominent feature is its ability to combine multiple images into a single picture. For example, a business could place a product photo into different backgrounds or merge design elements into a room layout without needing manual editing.
“When we first launched native image generation in Gemini 2.0 Flash earlier this year, you told us you loved its low latency, cost-effectiveness, and ease of use. But you also gave us feedback that you needed higher-quality images and more powerful creative control,” the tech titan noted.
For developers, Gemini 2.5 Flash Image arrives fully integrated across Google’s ecosystem. It is available through the Gemini API, within Google AI Studio, and in Vertex AI for enterprise-scale use. Also, Google AI Studio now offers ‘build mode’ templates for rapid experimentation, enabling everything from custom image filters to mockup generators to be prototyped with simple natural-language instructions.
The model is priced at $30.00 per one million output tokens, with each generated image averaging around 1,290 output tokens, which works out to around $0.039 per image. The search giant is also rolling out Gemini 2.5 Flash Image through a partnership with Adobe. Currently, Adobe Firefly users can create up to 20 images for free, while Creative Cloud Pro subscribers get unlimited access. And from September 1, the model will also be available in Adobe Express.
At the same time, to support responsible AI use, every image made with Gemini 2.5 Flash Image includes a hidden SynthID watermark. Notably, this mark can not be seen by the human eye but can be detected to show that the picture was created with AI. The development comes at a time when competition in AI image generation is intensifying across the tech industry. For example, in April 2025, OpenAI extended its GPT-4o-powered image generation tools to free-tier users, after initially limiting them to paid subscribers, giving a wider audience access to its creative AI features. The company is also expanding into video with Sora, a system capable of generating realistic moving scenes. Recently, Meta (after slower progress with its own tools) partnered with Midjourney to bring high-quality image generation into Facebook, Instagram, and WhatsApp.
