Google stimulates Gemini Ai image possibilities in the last salvo against Chatgpt

Google launched Gemini 2.5 Flash image on Tuesday and delivers a new AI model that generates and edits images with more precision and character consistency than earlier tools – which would close the gap with OpenAi’s Chatgpt.

The urge of the technical giant to integrate advanced image editing into Gemini reflects a wider push between AI platforms to absorb image generation as a must-have function. With the new tool, now available on Gemini apps and platforms, users can edit visuals using natural language complex tasks such as pose changes or multi-image merger without distorting faces or scenes.

In a blog post, Google said that the model enables users to place the same character in different environments, [and] Show a single product from multiple corners … all while retaining the subject. “

The model first appeared under the pseudonym “Nano-Banana” on Crowdsourced test site Lmarena, where it attracted attention for his seamless operation. Google confirmed on Tuesday that it was behind the tool.

Google said that the system can merge multiple images, keep character consistency for telling stories or branding and integrate ‘world knowledge’ to interpret diagrams or combine reference material – all within a single prompt.

The model costs $ 30 per million output tokens – about four cents per image – on Google Cloud. It is also distributed via OpenRouter and Fal.ai.

OpenAI introduced the GPT-4O model in May 2024 and added image generation in March 2025, which helped with the use of the use of chatgpt above 700 million weekly active users. Google reported 400 million monthly active Gemini users in August 2025, which would indicate a weekly use that considerably openly paths.

Google said that all outputs will include an invisible synthid watermark and metadata tag to mark them as AI-generated to tackle worries about abuse and authenticity.