Overview
Nano Banana Pro is Google DeepMind’s premium image generation model, released in November 2025. Built on Gemini 3 Pro, it’s a significant step up from the original Nano Banana, bringing 4K resolution, advanced prompt-driven image editing, support for up to 4 reference images, and substantially improved text rendering across multiple languages.
Where the original Nano Banana prioritizes speed and affordability, Nano Banana Pro prioritizes quality and control. It’s designed for professional asset production, using advanced reasoning to follow complex instructions and render high-fidelity detail. If you need the output to be production-ready rather than a starting point, this is the model to reach for.
Getting Started
- Go to Image Generation — Navigate to krea.ai/image and select the model from the dropdown.
- Select Nano Banana Pro — Open the model picker and choose Nano Banana Pro from the Intelligent Models section.
- Write a detailed prompt — Pro rewards specificity. Include subject, style, lighting, composition, mood, and any text you need rendered.
- Add reference images (optional) — Upload up to 4 reference images to guide style, composition, or subject identity.
- Choose your resolution and aspect ratio — Pro supports up to 4K. Start at a lower resolution for drafts, move to 4K for finals.
- Generate — Click Generate. Pro is slower than the original Nano Banana, but the output reflects the added processing.
- Edit and refine — Use prompt-driven editing to make targeted changes to your result without starting from scratch.
At a Glance
| Feature | Detail |
|---|
| Speed | Medium (2/3) |
| Credits | ~100 per generation |
| Underlying model | Gemini 3 Pro Image (Google DeepMind) |
| Resolution | Up to 4K |
| Reference image support | Up to 4 images |
| Image editing | Yes — prompt-driven |
| Text rendering | Excellent — multi-language |
| Best at | Complex tasks, professional output, image editing |
Overview
Nano Banana Pro brings advanced reasoning to image generation. It interprets complex, multi-condition prompts more accurately than most models, understanding not just what you’re asking for, but how the elements in your scene should relate to each other. That makes it reliable for outputs that need to be right the first time rather than iterated toward.
Its text rendering is one of its strongest features. Multi-language typography, dense layouts, brand logos, and product labels are all handled with high accuracy, making it a practical choice for design-oriented work where text is central to the visual.
The prompt-driven editing capability is also worth highlighting. Rather than regenerating an image from scratch when you want a change, you can describe the specific edit you want and the model will apply it while preserving everything else: lighting, composition, facial likeness, and color tone remain consistent across edits.
When to Use Nano Banana Pro
| Use When | Avoid When |
|---|
| You need final-quality, production-ready output | You’re in an early drafting phase and need speed |
| Your image requires legible, stylized text | You’re on a tight credit budget |
| You need to edit an existing image with precision | You need more than 4 reference images |
| Your prompt is complex and requires accurate interpretation | A faster model would serve the task just as well |
| You need 4K resolution for print or large-format output | |
Common Use Cases
- Professional asset production: Marketing visuals, campaign imagery, brand collateral
- Text-heavy designs: Posters, product labels, signage, advertising layouts
- Character and portrait work: High-fidelity human figures with consistent likeness across edits
- Architectural and product visualization: Photorealistic renders with fine spatial detail
- Image editing: Targeted modifications to existing images with composition preservation
Prompting Tips
| Tip | Example |
|---|
| Front-load the most important information | Start with the subject and core visual before adding detail |
| Describe text content in quotes | A poster reading “Grand Opening — March 2026” |
| Use reference images to define visual identity | Upload brand assets, style references, or character images |
| Specify lighting precisely | ”Rim lighting from the left, soft fill from front, dark background” |
| Use the editing capability for targeted changes | ”Change the jacket color to navy, keep everything else the same” |
| Start at 2K for review, output at 4K for delivery | Keeps credit costs manageable during iteration |
Example Prompts
A high-fashion editorial set in a Tokyo convenience store at 2am, model in an oversized iridescent coat, fluorescent lighting, rain-streaked window in background, Comme des Garçons aesthetic, film grain
A bold typographic poster reading "The Future Is Already Here" in a heavy condensed sans-serif, distressed concrete texture background, red and white color palette, large format print quality
A cinematic still from a fictional 1970s spy thriller, a woman in a fur coat stepping out of a black car on a rain-slicked Paris street at night, neon reflections on the pavement, film grain, anamorphic lens flare
A premium olive oil brand packaging mockup, the label reads "Groves of Crete • Cold Pressed Extra Virgin," hand-drawn botanical illustration of olive branches, matte cream label on a dark glass bottle, shot on linen background
