If you have seen the 3D figurine photos flooding Instagram and X lately — real people's selfies transformed into hyper-realistic miniature toy figures — there is a good chance Nano Banana made them. And if you have not heard of Nano Banana yet, that changes now.
Nano Banana is Google DeepMind's family of AI image generation and editing models, built on the Gemini architecture. The name sounds playful — and it was. It originated as a secret codename when Google was quietly testing the model on LMArena, the crowd-sourced AI evaluation platform, in August 2025. Users did not know what they were rating. They just knew the anonymous model was extraordinary. The name stuck, the internet adopted it, and when Google officially revealed its identity as Gemini 2.5 Flash Image, it embraced the Nano Banana brand entirely. Google later confirmed the nickname originated from nicknames given to a Product Manager at Google DeepMind named Naina Raisinghani.
When it launched publicly on August 26, 2025, it attracted over 10 million new users to the Gemini app and facilitated more than 200 million image edits within weeks. No AI image tool had scaled that fast.
The three models in the Nano Banana family
There are now three distinct Nano Banana models, each with a different speed and quality trade-off.
The original Nano Banana, officially Gemini 2.5 Flash Image, launched in August 2025. It is fast, accessible, and handles the core image editing and generation tasks that made the product viral — hairstyle changes, background swaps, multi-image fusion, and the 3D figurine transformation that became the defining use case of its launch.
Nano Banana Pro, officially Gemini 3 Pro Image, launched November 20, 2025. Built on Gemini 3 Pro, it adds the reasoning and world knowledge of Google's most advanced language model to image generation. It connects to Google Search for real-time context — so when you ask it to generate an accurate map, a scientific diagram, or a product infographic that gets the facts right, it can pull from real-world knowledge rather than hallucinating details. It supports up to 4K resolution and handles text rendering in multiple languages, including accurate translation of text within images. Nano Banana Pro is now integrated into Google Ads, Workspace (Slides and Vids), Vertex AI, Adobe Firefly, Canva, Figma, and Photoroom.
Nano Banana 2, officially Gemini 3.1 Flash Image, launched February 26, 2026. It combines the quality of Nano Banana Pro with the speed of Gemini Flash. As of February 27, 2026, it became the default image model across the Gemini app's free and paid plans, as well as Google Search (AI mode), Lens, AI Studio, Vertex AI, and Google's Flow video editing suite. For the first time, professional-grade image generation is free to the public.
What makes these models different
The capabilities that set Nano Banana apart from competitors come down to four things. First, subject consistency: Nano Banana 2 can maintain the appearance of up to five characters and the fidelity of up to 14 objects across a single editing workflow — essential for storyboards, comics, or any project where the same character needs to appear across multiple scenes without drifting in appearance.
Second, text rendering: generating legible, accurate text inside images has been the Achilles heel of AI image models for years. Nano Banana 2 renders precise typography directly in images, supports multiple languages, and can translate text within an existing image — a genuine breakthrough for global marketing and localization workflows.
Third, real-world grounding: Nano Banana Pro and Nano Banana 2 connect to Google Search, meaning they can generate accurate visuals of real-world subjects, current events, and factual content. This is what enables accurate infographics, recipe visualizations, and data diagrams that get the details right.
Fourth, SynthID watermarking: every Nano Banana output carries an invisible SynthID digital watermark, allowing verification of AI-generated origin. This is Google's transparency standard across all its AI-generated media.
How to use Nano Banana right now
The simplest path is the Gemini app. Open it, select the banana emoji Create images tool from the menu, and start prompting. Free users get the Nano Banana 2 (Fast) model. Google AI Plus, Pro, and Ultra subscribers get higher quotas and access to Pro mode, which uses Nano Banana Pro for maximum quality. In Google AI Studio, developers get API access for building applications on top of any Nano Banana model. In Vertex AI, enterprises can access Nano Banana Pro for scaled creative production.
The formula that gets the best results: start with a clear subject, add an action, describe the scene and lighting, then specify any style or format requirements. The more specific the prompt, the more precisely the model follows it.