Google's Nano Banana 2 Launching Soon, Likely To Be Powered By Gemini 3 Pro Visual AI

Google is gearing up for the launch of Nano Banana 2, its next-generation AI image editing tool, which claims to deliver better accuracy and creativity in visual content creation. The new AI tool will be powered by the advanced Gemini 3 Pro visual AI large language model. It is expected to roll out as early as this week, following leaks that have ignited excitement across social media platforms.

The original Nano Banana, introduced earlier this year as part of Google's Gemini 2.5 Flash rollout, made waves with its ability to handle complex image edits and generate visuals from natural language prompts. However, Nano Banana 2 aims to surpass its predecessor by tackling longstanding limitations in text rendering, infographics, charts, and adherence to intricate user instructions – areas where rival models have often faltered.

Nano Banana 2: Expected new features

Leaked details, shared by tech tipsters reveal that the new model will support a wider array of output resolutions, including 1K, 2K, and 4K, alongside expanded aspect ratios like 9:16 and 16:9. This upgrade directly addresses user complaints about inconsistent image dimensions in the first version.

One viral post showcased a surreal composition featuring US President Donald Trump, Ukraine's Volodymyr Zelenskyy, and Russia's Vladimir Putin displayed on a vintage monitor, highlighting the model's creative flexibility. Speculation is rife that Nano Banana 2 draws power from Google's forthcoming Gemini 3.0, which unifies multimodal reasoning across text, vision, and generation for more cohesive outputs.

Beyond resolution boosts, the tool introduces a multi-step generation process that allows it to plan, evaluate, and refine images autonomously. It will spot discrepancies in its own outputs and make corrections, enhancing realism and detail fidelity. The Gemini 3 Pro backbone is tipped to excel in rendering fine text and visual elements, making it ideal for crisp posters, design graphics, and professional exports with native 2K renderings and 4K upscaling.

Improved cultural awareness is another highlight, with the model trained on diverse regional datasets to produce geographically authentic scenes. For instance, prompts like "a family picnic in Tokyo springtime" or "streetwear shoot in Berlin winter" are expected to yield visuals that capture local nuances accurately. Character consistency will also see refinements, enabling seamless sequences of images featuring the same subjects.

Google is reportedly experimenting with "Edit with Gemini", a feature that lets users highlight and modify specific photo elements – such as backgrounds, clothing, or lighting – without regenerating entire images. This could streamline workflows for designers and content creators.

The frenzy around Nano Banana 2 has been amplified by its brief appearance on internal Google platforms before being pulled down, fuelling anticipation for a swift public debut. As AI image tools become staples in creative industries, Nano Banana 2 positions Google as a frontrunner in delivering intuitive, high-fidelity generation that blends advanced reasoning with user-friendly editing.

Published on: Wednesday, November 12, 2025, 02:59 PM IST