Nano Banana 2
Snappy, high-impact stills with strong style control — built for social, branding, and bold creative concepts. Use Nano Banana 2 in the same GPT Image 2 workflow: fast iteration, on-brand stills, and crystal-clear in-frame text when you need it.
What is Nano Banana 2?
Nano Banana 2 (Gemini 3.1 Flash Image) is Google DeepMind's latest production-grade image model, launched February 2026. It brings Flash-tier speed to advanced generation — with accurate real-world depictions powered by Google's knowledge base, best-in-class multilingual text rendering including in-image translation, and subject consistency across up to 5 characters and 14 objects, all at roughly half the cost of Nano Banana Pro.
Knows Real Places, Products & Culture
Backed by Google's world knowledge, Nano Banana 2 accurately depicts real locations, landmarks, products, and culturally authentic settings — without you needing to upload any reference images.
Multilingual Text & In-Image Translation
Render sharp, legible in-image text across languages and translate embedded copy — English to Spanish, Hindi, Arabic, and more — without altering the surrounding composition.
Flash-Speed at Half the Cost
Standard images in 4–6 seconds, 4K at ~$0.15 — approximately 50% cheaper than Nano Banana Pro. Supports 14 aspect ratios including 4 ultra-wide options exclusive to NB2.
How to Use Nano Banana 2
Explore how Nano Banana 2's Image Search grounding, multilingual text rendering, subject consistency, and Flash-speed generation fit into real production workflows.
Image Search grounding for real-world accuracy
Google Image Search · Real location & product reference · NB2 exclusive
Generate Grounded Visuals with Google Image Search
Nano Banana 2 is the only model in the Nano Banana family with access to Google's Image Search during generation. When your prompt references a real landmark, product, or recognizable subject, NB2 actively searches for visual references to anchor the output — so you get accurate architecture, true-to-life products, and faithful depictions of real-world scenes without uploading reference images yourself. Essential for travel content, product documentation, and any creative work where factual accuracy matters.
Try now
Precision text rendering & multilingual translation
Legible in-image type · Translate text within images · Global campaigns
Render & Translate In-Image Text with Precision
Nano Banana 2 delivers best-in-class in-image typography — sharp, legible text at any size across posters, product labels, social cards, and marketing mockups. More uniquely, it can translate text already embedded in an image — converting English copy into Spanish, Hindi, Arabic, and other languages without altering the surrounding visual composition. One prompt, one image, localized for every market. Ideal for global campaigns where producing localized creative at scale would otherwise require a full design pass per language.
Try now
Subject consistency across series & storyboards
Up to 5 characters · 14 reference objects · Narrative continuity
Maintain Character & Object Consistency Across a Series
Nano Banana 2 preserves the visual identity of up to 5 characters and 14 distinct objects within a single workflow — keeping faces, outfits, product shapes, and brand elements consistent across every frame without retraining or additional fine-tuning. Use it to build storyboards, multi-scene ad sequences, and product catalogs where the same cast and props must appear reliably from shot to shot. What used to require hours of manual consistency checking now ships in minutes.
Try now
Flash-speed iteration at half the cost
4–6s generation · 512px to 4K · 14 aspect ratios incl. ultra-wide
Iterate at Flash Speed Across Any Format or Resolution
Built on Gemini 3.1 Flash Image, Nano Banana 2 generates standard-resolution images in 4–6 seconds and scales to 4K — at roughly half the cost of Nano Banana Pro. It supports 14 aspect ratios including four ultra-wide options not available in any other Nano Banana model, covering everything from vertical social posts to widescreen backdrops. Use it to run high-volume A/B creative tests, spin up seasonal variants, and keep production pipelines moving without burning budget on render time.
Try now
Why Creators Choose Nano Banana 2
Google's Gemini 3.1 Flash Image brings unique capabilities no other model offers — from live Image Search grounding to in-image multilingual translation — at Flash speed and half the cost of Pro.
Knows Real Places, Products & Culture
Nano Banana 2 draws on Google's world knowledge and image search to accurately depict real locations, landmarks, products, and culturally authentic settings. Describe a Tokyo street, a specific shoe model, or a local festival — and get it right, without uploading a single reference image.
In-Image Text Translation
Translate text embedded inside an image — English to Spanish, Hindi, Arabic, and more — without altering the surrounding visual composition. One generation, localized for every market.
Same Face & Style, Every Single Shot
Keep the same face, hairstyle, expressions, and clothing for up to 5 characters — and the exact look of up to 14 products or objects — across an entire series. Build storyboards, campaign sets, and product catalogs where every image unmistakably belongs to the same world.
Flash-Tier Speed
Standard images in 4–6 seconds, backed by Gemini 3.1 Flash architecture. Rapid iteration across concepts, formats, and localized variants without waiting on slower Pro-tier render times.
14 Aspect Ratios incl. Ultra-Wide
Supports 14 aspect ratios — including four ultra-wide options exclusive to NB2 that aren't available in Nano Banana or Nano Banana Pro. From vertical Stories to widescreen backdrops in a single workflow.
~50% Lower Cost Than Pro
4K output at approximately $0.15 per image — roughly half the price of Nano Banana Pro. Scale high-volume A/B testing, batch catalog generation, and seasonal variant production without doubling the budget.
Create with Nano Banana 2 in 3 Simple Steps
From brief to production-ready visual in seconds. Nano Banana 2's Google Image Search grounding, multilingual text rendering, and Flash-speed generation make the whole workflow faster.
Step 1: Describe Your Subject or Upload References
Describe your scene, subject, mood, and any text you want embedded. For real-world places, products, or culturally specific settings, Nano Banana 2 draws on Google's world knowledge to get the details right — no reference image required. Upload your own references to lock a character's face, outfit, or product appearance across a series.
Step 2: Generate & Localize in Seconds
Nano Banana 2 produces images in 4–6 seconds at your chosen aspect ratio (including ultra-wide) and resolution up to 4K. Need the same image in multiple languages? Prompt the model to translate in-image text — it localizes the copy without touching the surrounding visual.
Step 3: Download & Scale
Download at full resolution and deploy immediately — or batch additional variants for A/B testing, seasonal updates, or market-specific localizations. At ~$0.15 per 4K image, scaling a full campaign set costs a fraction of traditional production.
What the Community Says About Nano Banana 2
Creators, global marketing teams, and brand designers share how Nano Banana 2 changed their production workflow.
"The Image Search grounding is a game-changer. I described a specific street in Tokyo and Nano Banana 2 actually looked it up and got the architecture right. No reference image needed — it just knew."
"We run campaigns across 8 markets. With Nano Banana 2, I generate the hero image once and ask it to translate the embedded text into each language. The layout stays intact — only the copy changes. It cut our localization time by 70%."
"Subject consistency across 14 objects in one workflow is genuinely impressive. I feed it my product lineup and every image in the series shows the same packaging, same colors, same shapes — without fine-tuning."
Frequently Asked Questions
Got questions? We've got answers.