Nano Banana 2.1: Google's Updated Image Model
Nano Banana 2.1 is Google's image generation model, generally available since October 6, 2026. Google describes it as an update to Nano Banana 2 with better visual quality, character consistency and text rendering. In Slates it costs 3 credits per image at 1K, 4 at 2K and 8 at 4K, and it takes up to 14 reference images.
| Resolution | Credits per image |
|---|---|
| 1K | 3 |
| 2K | 4 |
| 4K | 8 |
Slates credits per image, checked against the Slates model registry on every build. The exact cost shows before you generate.
What Nano Banana 2.1 is
Nano Banana 2.1 is Google's newest Nano Banana image model. Google made it generally available on October 6, 2026, and describes it as an update to Nano Banana 2 with improved visual quality, character consistency and text rendering.
It is a language model that outputs pixels. It reads your whole prompt and reasons about it before it draws, which is why it rewards plain descriptive sentences over stacked keywords.
In Slates it is the one Nano Banana model. Nano Banana 2, Nano Banana 2 Lite and Nano Banana Pro left the model picker on October 23, 2026, and every image you already made on them keeps its original label.
One naming note. There is a third-party desktop app called "Nano Banana Pro", named after an older Google model. The app and the model are unrelated products.
What it costs in Slates credits
Nano Banana 2.1 has three resolution tiers, and the price rises with the pixel count.
In Slates it is 3 credits per image at 1K, 4 at 2K and 8 at 4K. An edit with up to 14 reference images costs the same as a fresh generation at that resolution. The exact cost shows on the Generate button before you commit.
There is no Google account to set up and no key to manage. You buy credits once and spend them across every model in the app, with no monthly expiry on what you've already paid for. The full pack range is on the pricing page.
Where Nano Banana 2.1 earns its place
Slates opens on GPT Image 2.5 as its default image model, because its frames tend to look more cinematic, with more natural light. Nano Banana 2.1 is the model you switch to for a different kind of result.
Text inside the image is its first strength. Google names text rendering as one of the upgrades, so signs, labels, posters and title cards are where it is worth a try first. Put the exact words in quotes and say what kind of type they should be set in.
Character work is the second. Up to 14 reference images fit in one request, which gives it room for several characters, a location and a product in the same frame. Clear, front-facing reference photos help it most. That makes it a strong base for an AI influencer or a recurring cast in an AI story.
Third, it is a clean second opinion. When a frame from the default model comes back with grainy, patterned noise, rerolling the same prompt on Nano Banana 2.1 often fixes it. But its own failure runs the other way: skin can look too smooth and fine detail can go soft. Both models can produce fully photoreal frames, so judge the frame, not the model.
How to prompt it
Describe the scene in sentences. Name the subject, what it is doing, where it is, how the shot is framed and the style you want, in roughly that order.
Say what you want, not what you don't. "An empty street at dawn" works better than "a street with no cars".
Use camera language for framing: low angle, close-up, aerial view. In an edit, say what must stay the same as well as what changes.
Lighting needs a firm hand. Asked for a single cold light on one side of a face, Nano Banana 2.1 has lit the face warm and from the front instead. State the light source early and describe the exposure you want to see, such as one half of the face falling to near black.
Google lists three known limits: very small text, imperfect character consistency and left-right confusion. Keep rendered type large and short, test a reference setup before a large batch, and when it matters say whose left and right you mean, since the model reads them from the viewer's side.
When to pick another model
Google's safety filters can refuse realistic people in some poses, fashion editorial work and commercial-photography setups. When a prompt is refused, FLUX 3 Image and Seedream 5 Lite are the less filtered image models in Slates, and you can switch with one click and keep the same prompt.
Frequently asked questions
What is Nano Banana 2.1?+
Nano Banana 2.1 is Google's image generation model, generally available since October 6, 2026. Google describes it as an update to Nano Banana 2 with improved visual quality, character consistency and text rendering. It reasons about a prompt before drawing, so it responds best to full descriptive sentences.
How much does Nano Banana 2.1 cost per image?+
In Slates, Nano Banana 2.1 is 3 credits per image at 1K, 4 at 2K and 8 at 4K. Edits cost the same as a generation at the same resolution. The exact cost shows on the Generate button before you commit, and no Google account is required.
What happened to Nano Banana 2 and Nano Banana Pro in Slates?+
Slates replaced Nano Banana 2, Nano Banana 2 Lite and Nano Banana Pro with Nano Banana 2.1 on October 23, 2026. Images you already made on the older models keep their original labels, and new generations run on 2.1.
Is Nano Banana 2.1 the same as the Nano Banana Pro app?+
No. Nano Banana 2.1 is an image model made by Google. Nano Banana Pro is also the name of a separate third-party desktop app, named after an older Google model. The two share a name and nothing else.
Can Nano Banana 2.1 generate realistic people?+
Yes, and it can look fully photoreal, though skin sometimes comes out too smooth. Google's safety filters can refuse some poses, fashion editorial work and commercial-photography setups. For those prompts, FLUX 3 Image and Seedream 5 Lite are the less filtered image models in Slates.
Related
Run Nano Banana 2.1 in a real workflow
Slates handles the generation, the file management, the @mention character system, and the timeline for video work that starts from your Nano Banana stills.
Get Slates