Image: Google

Update: Google has now made it official. "Meet Nano Banana 2.1, our latest image generation and editing model," the company posted, confirming the name people spotted in Flow. Google says it outperforms its previous image models across the board, with big leaps in visual design, mask-based editing and keeping subjects consistent from image to image.

Meet Nano Banana 2.1, our latest image generation and editing model 🍌

— Google (@Google) October 6, 2026

According to TestingCatalog, the model is rolling out in Google AI Studio, Google's API and the Gemini app. Google describes it as a high-quality image model "at Flash scale." Image output costs $30 per million tokens, which works out to about $0.034 for a 1K image, $0.050 for 2K and $0.076 for 4K. Based on that pricing, TestingCatalog suspects it may really be a Pro-tier model.

GOOGLE 🔥: Nano Banana 2.1 is rolling out on Google AI Studio, APIs, and Gemini!

— 🚨 AI News | TestingCatalog (@testingcatalog) October 6, 2026

---

Nano Banana 2.1, the next version of Google's popular image model, is showing up inside Google Flow. TestingCatalog, which tracks unreleased features in Google's apps, says the model "appears to be available" in Flow right now. Several other users have posted screenshots of it.

Google hasn't officially announced Nano Banana 2.1 yet. There's no blog post or model page, and nothing on Google's official X accounts so far.

GOOGLE 🔥: Nano Banana 2.1 appears to be available on Google Flow right now.

Yet its output seems similar to what Nano Banana Pro generates, so it is unclear whether requests are actually routed to a new model. https://t.co/2aikpVWcO2 pic.twitter.com/STtLs9xwOa

— 🚨 AI News | TestingCatalog (@testingcatalog) October 5, 2026

TestingCatalog added an important caveat. Its early outputs look a lot like what Nano Banana Pro produces, so it isn't clear yet whether requests are really going to a new model.

One of the earliest sightings came from AiBattle, whose screenshot TestingCatalog quoted. It shows Flow's image model menu with three choices: Nano Banana Pro, Nano Banana 2 Lite and Nano Banana 2.1.

Nano Banana 2.1 is live on Google Flow pic.twitter.com/HvXHfKmdkK

— AiBattle (@AiBattle_) October 5, 2026

That menu is the most telling detail, because Nano Banana 2 is missing from it. When Google launched Nano Banana 2, it made it the default image model in Flow, free for all users at zero credits.

Now 2.1 sits in that same spot. Another user's screenshot shows it selected with a "Generating will use 0 credits" note. That looks like a swap of Flow's free default model, not a new extra for paying users.

It also completes a trail we reported on earlier, when a Flow web build renamed a "Nano Banana 2.5 Flash" reference to "Nano Banana 2.1." The name has now moved from hidden code to a menu people can actually click.

Flow is a natural place for this to happen first. Google introduced Flow as an AI filmmaking tool built around its Veo, Imagen and Gemini models, and it grew out of a Google Labs experiment.

Recent Nano Banana launches have put Flow near the front of the line. Nano Banana Pro started rolling out to Flow for Ultra subscribers on day one, and Google listed Flow among the apps getting Nano Banana 2 Lite. A Labs tool full of creative power users is a lower-risk test bed than the Gemini app, which reaches far more people.

Early testers are already comparing results. AJ ran the same prompt and settings through Nano Banana 2 Lite and 2.1 in Flow. His verdict: the images are very different, and viewers can be the judge.

Google next image generation model Nano Banana 2.1 is now available in Google Flow.

Here's how it compares to Nano Banana 2-lite, both same prompt and settings, generated on Flow !

First image: Nano Banana 2 lite
Second Image: Nano Banana 2.1

Things are very different for the… pic.twitter.com/hFQzqThvBA

— AJ (@ItsmeAjayKV) October 5, 2026

Others are pitting it against rival models. The account thehype ran five prompts through Nano Banana 2.1 in Flow and Tencent's HY Image 3.5, and called the result "closer than we expected." In its tests, Nano Banana kept characters closer to the reference images, while Tencent's model won on light and mood.

google's new nano banana 2.1 vs tencent's hy image 3.5 – closer than we expected

same five prompts, one shot each, no edits, 3d style

models: @TencentHunyuan hy image 3.5 via @OpenRouterAI, @GoogleDeepMind nano banana 2.1 in google flow

- nano banana keeps the characters… pic.twitter.com/EnAqTG1BJu

— thehype. (@thehypedotnews) October 5, 2026

Treat those tests with care. A model menu is just a label in the interface. Which model actually answers a request is decided on Google's servers, and nothing on screen proves it.

If TestingCatalog is right that outputs look like Nano Banana Pro, there are two simple explanations. The new name may point to an existing model while the rollout finishes. Or 2.1 may simply be close to Pro in quality, which is exactly how Google pitched Nano Banana 2: Pro capabilities at Flash speed.

The version number sets expectations, too. Going from 2 to 2.1 points to a refinement of Nano Banana 2 (officially Gemini 3.1 Flash Image), not a new generation. The gains are more likely to be steady than dramatic.

What to watch next: an official post on Google's blog, a new model ID in the Gemini API and AI Studio, and Nano Banana 2.1 showing up in the Gemini app. Until then, Flow users may be able to try it, but Google hasn't confirmed what it is.