Google's image model, explained
Nano banana is the internet's nickname for Google's image generation and editing model, the one built into the Gemini app. The name started as a codename and went so viral that Google kept it. The current version, nano banana 2.1, shipped on October 6, 2026, and it sits behind the flood of "edit any photo with a sentence" clips filling social media.
This page explains what it actually does, how to use it step by step in the Gemini app, what it's genuinely good at — and the one job it doesn't do, which is the reason tools like ours exist.
Animate a photo — $9.90 · 3 renders
Strip away the meme name and it's an image model with one specific talent: it edits a photo you give it, following plain-language instructions, while keeping the people in the photo recognizably themselves. Swap a background, change an outfit, add a ridiculous hat — Grandma still looks like Grandma. That ability is called character consistency, and it's the reason this particular model went viral when so many image editors didn't.
One boundary worth stating plainly: it's a still-image tool. It produces and edits pictures. It doesn't make videos, and none of its outputs move — no matter how the clips on your feed make it look.
Open the Gemini app (or gemini.google.com), start a chat, and attach the photo you want to edit. Then describe the change in plain words — "put her in a sunflower field at golden hour", "make it look like a vintage polaroid". The model returns an edited image; you keep instructing until it's right.
The free tier works, with usage limits — enough for casual editing, not for heavy daily use. The exact caps change often, so trust what the app tells you over any fixed number you read online, including here. Paid Google plans raise the ceilings; we won't quote figures that could be stale by next month.
Genuinely good at: consistent people across edits, multi-part changes described in one sentence, and fast iteration — you can try five versions of an idea in five minutes. For photo editing as a hobby or for social content, it's the default for a reason.
The hard boundary: it edits stills. If what you actually want is for the photo to move — a blink, a smile, a small dance — that is video, a different model category entirely. No amount of re-prompting an image model will make the picture wave back.
Here's the honest bridge, from the team behind a photo-to-video tool: if your goal ends at a better still image, stay in Gemini — it's excellent, and the free tier probably covers your needs. If your goal is the photo coming alive — an old portrait that smiles again, a baby photo that bounces — you need an image-to-video model, and a general chat app is the wrong shape for that job.
That's what PhotoAlive does, as a one-shot tool: upload one photo, pick a motion style from real samples, and get 5-second 720p videos in about two minutes — $9.90 for 3 renders, no account, automatic refund if generation fails. If you're curious what "coming alive" means in practice, our guide to animating old photos shows exactly what changes. Use whatever fits the job — that's the whole point of knowing what nano banana is.
The viral nickname for Google's image generation-and-editing model in the Gemini app. It began as a codename, spread as a meme, and Google embraced it. The current version, nano banana 2.1, shipped on October 6, 2026.
There is a free tier in the Gemini app, with usage limits that change over time — we won't quote specific numbers that go stale. When you hit the ceiling, the app tells you; paid Google plans raise the limits.
Nano banana 2.1 shipped on October 6, 2026, and is the version of the image model rolling into the Gemini app now. The line's defining strength is unchanged: edits that keep the same person recognizably themselves. For a detailed changelog, Google's own announcement is the source of record.
No — it edits and generates still images. Making a photo move is image-to-video, a different category. That's the job PhotoAlive is built for: one photo in, a 5-second video out, $9.90 for 3 renders, no account.
Yes — character consistency is its headline ability and the reason it went viral: edits change the scene, not the person. Keeping a face consistent across still edits, though, is a different task from animating that face naturally in video.
No signup: the browser you pay from is your account. Buy renders once, use them any time — any photo, any style, even motion in your own words.
Animate a photo — $9.90 · 3 renders