Google DeepMind launches Nano Banana Pro: Gemini 3 Pro image model for accurate text and studio-grade visuals

Nano Banana Pro, also known as Gemini 3 Pro Image, is a new image generation and editing model built by Google DeepMind on Gemini 3 Pro. It is positioned as a state-of-the-art image creation and editing system that must respect structure, world knowledge, and text layout, not just style. Nano Banana Pro follows Nano Banana, which is based on Gemini 2.5 Flash Image and focuses on fast and casual image editing, such as restoring photos and generating statues.

From Gemini 2.5 Flash image to Gemini 3 Pro image

Early Nano Banana models were targeted at casual creators for quick creative editing. It helps restore old photos and build stylized 3D mini statues with simple tips. Nano Banana Pro retains the editing pipeline but runs on top of Gemini 3 Pro, bringing more powerful reasoning and real-world knowledge to the image stack.

The model can transform prototypes, data sheets, and handwritten notes into diagrams and infographics that reflect the underlying information, rather than just producing decorative art.

Reasoning guidance, search-grounded visual effects

A core design point of Nano Banana Pro is inference-guided generation. Using Gemini 3 Pro, models can consume text, structured content, and references, and then plan images as explanations for that content. Nano Banana Pro can also connect to Google Search, using the search index as a real-time knowledge source.

Clear text and multilingual layout

Text within images is a long-standing failure mode for many diffusion-based generators. Nano Banana Pro definitely solves this problem. Google says it’s the best model in the Gemini series, producing images with correctly rendered and clear text, whether it’s a short slogan or a full paragraph.

Gemini 3 Pro’s multilingual reasoning flows into the image model. Nano Banana Pro can render text in multiple languages ​​and can also translate text that already appears in products or posters. The document shows a beverage can with the English text translated into Korean, while the visual design and layout remain unchanged.

Studio-level control, consistency and upgrades

Nano Banana Pro exposes a set of controls for design and production workflows, rather than single-shot artistic prompts. In terms of composition, the model can use up to 14 input images and maintain consistency and similarity for up to 5 people in a workflow. This supports tasks such as combining reference photos into a single fashion editorial, converting sketches into product shots, or keeping the same actors in multiple scenes.

There are several families of controls listed in the Studio Controls section of the model page. Users can change camera angles and shot types, including wide-angle lenses, panoramas, and close-ups, while controlling depth of field and focusing on specific subjects in the image. Color and lighting can be adjusted, such as changing day to night, replacing volumetric lighting with bokeh, or applying strong chiaroscuro effects without losing subject identity.

Nano Banana Pro supports explicit upgrades. Google’s official blog says it can produce crisp visuals at 1k, 2k or 4k resolutions, and provides examples of progressive zoom operations that preserve detail and composition. The aspect ratio is also programmable. Tips can be converted between 1:1, 4:3, 16:9, etc. ratios and film formats while keeping the protagonist locked in place and adjusting only the background.

Main points

• Nano Banana Pro is Gemini 3 Pro Image, an upgraded image generation and editing model following Nano Banana, which was based on Gemini 2.5 Flash Image and optimized for higher quality and control.
• The model integrates Gemini 3 Pro reasoning and Google search fundamentals so it can transform factual content, documents and real-time data into infographics, recipes, flowcharts and other information-dense visuals.
• It provides powerful text rendering and multi-language support, producing clear typography in images and the ability to translate or localize existing image text while preserving layout and design.
• Nano Banana Pro supports up to 14 input images and maintains the likeness of up to 5 people, with studio-style controls for camera angles, depth of field, lighting, aspect ratio and upscaling to 1k, 2k and 4k resolutions.
• The model is being deployed across Gemini apps, AI Patterns in Search, NotebookLM, Google Ads, Workspace apps, Gemini API, Google AI Studio, Vertex AI, Antigravity, and Flow, with all outputs watermarked using SynthID plus layer-specific visible watermarks.

Nano Banana Pro positions Gemini 3 Pro Image as a production-oriented imaging system that links Gemini 3 Pro inference, Google search fundamentals, and structured controls for layout, text, and upgrades. It directly addresses long-standing issues with text rendering, multi-language localization, and theme consistency, while retaining SynthID and visible watermarks as the default source signals across layers and surfaces. This launch brings Google’s imaging stack one step closer to providing developers and enterprises with an API-first, integrated vision platform.

Check technical details. Please feel free to check out our GitHub page for tutorials, code, and notebooks. In addition, welcome to follow us twitter And don’t forget to join our 100k+ ML SubReddit and subscribe our newsletter. wait! Are you using Telegram? Now you can also join us via telegram.