Google Ships Nano Banana Pro, Its Gemini 3 Pro Image Model

Google has released Nano Banana Pro, officially named Gemini 3 Pro Image, a higher-fidelity image generation and editing model built on Gemini 3 Pro, now rolling out in paid preview through the Gemini API in Google AI Studio and Vertex AI for enterprises. The release follows the launch of Nano Banana (Gemini 2.5 Flash Image) a few months ago, which the community adopted for character consistency, photo restoration, and local edits on an infinite canvas.

The stakes here are competitive. Text-to-image generation has become a battleground between Google, OpenAI, and Adobe, and Google is positioning Gemini 3 Pro Image as a production-grade tool rather than a novelty — one that can handle the precision demands of advertising, education, and enterprise content workflows.

Professional controls and resolution

Google says the model gives developers control over image physics — lighting, camera, focus, color grading — and composition to produce professional-quality outputs. It supports 2K and 4K resolution, meeting the standards required for professional production pipelines.

The model is built for compositing complex commercial assets. Google says it can maintain consistent resemblance for up to five individuals, integrate six high-fidelity shots, or blend as many as fourteen standard inputs into a single polished advertisement combining product images, logos, and references. Google has published a demo app that pairs logos with products to create mockup designs.

Text rendering and localization

Google claims a significant leap over 2.5 Flash Image in text rendering, producing clear, accurate text integrated into generated images — a longstanding weakness of image models and a requirement for marketing collateral and educational content. In a described demo, the model spelled out food words using realistic ingredients to express each word's meaning.

The company also frames the model as a localization tool. Gemini 3 Pro Image grasps the semantic context of an image and can change the language of elements like menus, signs, or documents via image-to-image generation while preserving the original artistic style or layout. Google demonstrated translating a beverage campaign's English text into French. A comic book generator app in Google AI Studio lets users create multi-page comic books featuring themselves and a friend, with advanced text rendering and stylization.

Grounding with Google Search

Gemini 3 Pro Image connects a knowledge base Google describes as vast, producing more factual assets than previous image generation models. When enabled, grounding with Google Search connects the model to real-time web content for data-driven outputs. Google points to applications requiring precise representations, such as biological diagrams or historical maps, and offers a demo app that generates infographics on any topic.

Distribution across the ecosystem

Google is widening access beyond its own tools. In Google Antigravity, the company's new agentic development platform, coding agents can use the image generation capabilities to produce UI mockups for user review or create visual assets before implementing code. Adobe and Figma are integrating the model into their creative platforms — a notable endorsement from two of the largest design software vendors.

Provenance and pricing trade-offs

Every image created or edited with Gemini 3 Pro Image carries a SynthID digital watermark to denote its AI-generated or edited origin, addressing growing regulatory and public pressure for provenance in AI-generated media.

Google positions the two image models as complementary: Gemini 2.5 Flash Image for faster, lower-cost generation, and Gemini 3 Pro Image for higher quality at higher cost and latency. The company says the release incorporates user feedback gathered since Nano Banana's launch and states plainly: "we aren't stopping here."

Developers can start through the Gemini API in Google AI Studio, with Vertex AI available for enterprise use, supported by documentation, a prompt guide, a cookbook, and a developer forum.