Log inSign Up
Open Hedra

Nano Banana Pro

All models
Image modelGoogle

Overview

Nano Banana Pro is an image generation and editing model developed by Google, built on the Gemini 3 Pro architecture. It uses deep visual reasoning and Google Search grounding to interpret complex prompts and incorporate real-world information. The model is particularly effective at precise text rendering, generating accurate infographics, and blending multiple reference images into a single cohesive scene for structured content generation.

Best of Nano Banana Pro

A high-resolution 5504x3072 image generated by Nano Banana Pro, showing a man walking across a city street at night. The scene uses a thermal imaging style, with the man and nearby cars glowing in warm orange and red heat signatures, while the surrounding Japanese neon-lit buildings are cast in cool blue and purple tones.Thermal City Crossing — Nano Banana ProA close-up rear view of futuristic crimson space suit boots walking across a sandy desert dune, kicking up fine sand. A tattered white cloak flutters above the mechanical legs. Generated by Nano Banana Pro at a resolution of 5504x3072.Sci-Fi Boots Walking in Desert — Nano Banana ProA split-screen comparison image. The left side displays a grey and blue running shoe on a cluttered office desk. The right side shows a transformed, vibrant blue and orange sneaker floating in a dark studio setting with smoke and water splashes, generated by Nano Banana Pro at 5504x3072 resolution.Nano Banana Pro: Running Shoe Commercial TransformationA split-screen comparison image generated by Nano Banana Pro at 5504x3072 resolution. On the left, a young woman with wavy brown hair smiles in a dimly lit, casual selfie. On the right, the same woman is shown in a brightly lit, professional corporate headshot wearing a navy blue blazer.Nano Banana Pro: Casual Selfie and Professional HeadshotAn ultra-high-resolution 5504x3072 image generated by Nano Banana Pro, showing a field of glowing, translucent tulip-like flowers in purple, yellow, and teal. Tiny lights sparkle inside each bloom under a dusk sky. Three luminous butterflies, two teal and one pinkish-red, flutter above the field.Glowing Flowers and Butterflies — Nano Banana ProA split-screen image showing a blue and orange running shoe. On the left, the shoe sits on a messy wooden desk under dim light. On the right, the same shoe floats in a dynamic commercial setting with smoke, water splashes, and lens flares, generated by Nano Banana Pro at 5504x3072 resolution.Running Shoe Transformation — Nano Banana ProA high-resolution 5504x3072 image generated by Nano Banana Pro showing a man walking down a Tokyo street. The entire scene is rendered in a thermal imaging style, with the central figure and car tires glowing in hot orange, while the surrounding buildings appear in cold blue.Thermal Imaging of Man in Tokyo — Nano Banana ProA high-resolution 5504x3072 image depicting a traditional Turkish breakfast spread on a rustic wooden table, generated by the Nano Banana Pro model. The spread includes simit bread, a fresh vegetable salad topped with feta cheese, small bowls of olives, cheese, butter, a pan of fried eggs with sausage, and a glass of Turkish tea.Traditional Turkish Breakfast Spread — Nano Banana ProA wide-angle shot of a hooded warrior in leather armor, viewed from behind, standing in a stone archway. Outside, numerous hot air balloons float over a desert landscape at sunrise. This high-resolution 5504x3072 image was generated using the Nano Banana Pro model.Hooded Warrior Overlooking Hot Air Balloons — Nano Banana Pro

What is Nano Banana Pro best for?

Nano Banana Pro is Google's high-end image model for design-grade work where accuracy matters. Its signature strength is text rendering — Google calls it the best model for generating correctly spelled, legible text directly in an image, across multiple languages — which makes it a workhorse for posters, ads, menus, infographics, and product mockups. Because it can ground a generation in Google Search, it keeps real-world facts straight in maps, diagrams, and data visualizations. It outputs up to 2K and 4K, holds up to 5 people consistent across as many as 14 reference images, and applies Gemini 3's reasoning to composition, camera, and lighting. Reach for it whenever text, layout, or factual detail would be costly to fix by hand.

Who makes Nano Banana Pro, and when was it released?

Nano Banana Pro is made by Google DeepMind and was released on November 20, 2025. 'Nano Banana Pro' is the nickname for Gemini 3 Pro Image — an image generation-and-editing model built on Gemini 3 Pro.

How does Nano Banana Pro relate to Google's other image models?

It is the premium tier of Google's Nano Banana image family. It succeeded the original Nano Banana (Gemini 2.5 Flash Image, August 2025) — the viral, fast, 1K model with no editing or reference-image support — and was itself followed by Nano Banana 2 (Gemini 3.1 Flash Image, February 2026), a Flash-tier model that reaches roughly 95% of Pro's quality at about half the cost and 2–3× the speed. Use Pro when you need the highest fidelity — text, 4K, complex composition; drop to Nano Banana 2 for fast, cheaper everyday generation. Google's broader image lineup also includes Imagen 4.

What are the best prompting tips for Nano Banana Pro?

Treat it as a thinking model, not a keyword matcher: write like a creative director instead of stacking tag soup. Google's guidance is to cover subject, composition, action, setting, and style, and to be specific about camera and lighting — real lens and aperture details beat vague terms like 'cinematic.' For revisions, use conversational editing: if an image is about 80% right, ask for the one specific change rather than regenerating from scratch. When combining inputs, the community's reference formula is [reference images] + [relationship instruction] + [new scenario]. Google's prompting tips for Nano Banana Pro and ultimate prompting guide have the full playbook.

What can Nano Banana Pro do that most image models can't?

Two capabilities power users lean on. First, legible, correctly spelled text at scale — multilingual paragraphs, signage, and infographics that most diffusion models garble — so you can design a finished marketing asset in a single generation. Second, factual grounding through Google Search: ask for a real map, a labeled diagram, or a data chart and it pulls accurate details rather than inventing them. Pair those with structured (pseudo-code) prompts that define variables for its reasoning engine, and it behaves as much like an information-design tool as an image generator. Community prompt collections such as awesome-nanobanana-pro catalog the tricks.

Similar models

Nano BananaGoogleNano Banana 2GoogleDreamina 3.1ByteDanceFlux 1.1 ProBlack Forest LabsFlux 1.1 UltraBlack Forest Labs

Prompt tips

  • Leverage multiple references: Upload up to 14 images to dictate style, character consistency, and environment simultaneously.

  • Prompt for precise typography: Explicitly state the exact text, font style, and placement you want; the model handles multilingual text and complex layouts reliably.

  • Use real-world data: When asking for infographics or charts, instruct the model to use its search grounding to pull accurate, up-to-date facts.

  • Think in 3D space: Describe the camera angle, spatial relationships, and lighting (e.g., "perfectly isometric" or "shot from below") to take advantage of its deep spatial understanding.