Nano Banana 2.1
Google's newest Gemini image model makes sharper text, cleaner infographics and more consistent people across edits. Here is what changed, what it costs and how it compares with Pro.
How to try it
Where to use Nano Banana 2.1
2.1 is a Google model; these are the official ways in.
Gemini app
Google lists the Gemini app among the products that use Nano Banana 2.1.
Google AI Studio or the Gemini API
Call gemini-nano-banana-2.1. Batch requests cost half the standard price; there is no free tier for this model.
Prefer a no-setup image tool?
LazyKiwi's image generator runs Nano Banana and other image models in one place, no API key needed.
What's new
What Nano Banana 2.1 improves
Google released Gemini Nano Banana 2.1 on October 6, 2026 as the successor to Nano Banana 2 (Gemini 3.1 Flash Image). The model card says it is based on Gemini 3.6 Flash.
Text and layout
Text, infographics and design
Google lists more accurate text rendering and infographic layout as a headline change. In Google's own human ratings, 2.1 with thinking scores 1048 for infographic design against 961 for Nano Banana 2 and 912 for Pro.
- More legible text in images
- Better infographic structure
- Thinking level: minimal, medium or high
Resolution
1K, 2K and 4K output
2.1 renders at 1K (default), 2K or 4K, and fixes stitching artefacts that wide and tall panoramas (1:4, 4:1, 1:8, 8:1) showed at 2K and 4K. It does not offer the 0.5K size that Nano Banana 2 has.
- 14 aspect ratios, from 1:8 to 8:1
- Up to 4096 × 4096 at 1:1
- No 512 px size
Consistency
Up to 14 reference images
One request can take up to 14 images: up to 4 people kept consistent and up to 10 objects kept as they are. Google's ratings put 2.1 ahead of both Nano Banana 2 and Pro on single- and multi-character consistency.
- Up to 4 characters
- Up to 10 objects
- Mask and scribble editing
Limits
What it still gets wrong
The model card notes that small text can blur, especially at 1K; long paragraphs render poorly; likeness between input and output is not always perfect; and left/right placement is sometimes swapped. Every image carries a SynthID watermark.
- Small text at 1K
- Occasional spatial mistakes
- SynthID on every image
Specs
Nano Banana 2.1 specs at a glance
From Google's API documentation and model card, October 2026.
- Developer
- Google (Gemini)
- API model id
- gemini-nano-banana-2.1
- Released
- October 6, 2026 (generally available)
- Based on
- Gemini 3.6 Flash
- Inputs
- Text, images (up to 14), video, PDF
- Output sizes
- 1K, 2K, 4K; 14 aspect ratios
- API price
- $1.50 / 1M input tokens; about $0.034 (1K), $0.050 (2K), $0.113 (4K) per image
- Where
- Gemini app, Google AI Studio, Gemini API, Vertex (Gemini Enterprise Agent Platform)
- In LazyKiwi
- Not available; the image generator offers Nano Banana
2.1 vs Pro
Nano Banana 2.1 vs Nano Banana 2 vs Pro
Elo scores are Google's own human ratings from the 2.1 model card (2.1 with thinking). Prices are Gemini API list prices per image.
| Model | Overall preference | Strongest at | Price per 1K / 4K image |
|---|---|---|---|
| Nano Banana 2.1 | 1050 | Text, infographics, multi-character consistency | $0.034 / $0.113 |
| Nano Banana 2 | 990 | Has a 0.5K size; now deprecated | $0.067 / $0.151 |
| Nano Banana Pro | 935 | World knowledge, brand consistency, up to 3 style references | $0.134 / $0.24 |
FAQ
Nano Banana 2.1 FAQ
What is Nano Banana 2.1?
Gemini Nano Banana 2.1 (model id gemini-nano-banana-2.1) is Google's image generation and editing model released on October 6, 2026. It replaces Nano Banana 2 and is based on Gemini 3.6 Flash, according to Google's model card.
Is "Gemini Nano Banana 2.1" the same model?
Yes. Gemini Nano Banana 2.1 is the official name; Nano Banana 2.1 is the short form.
Nano Banana 2.1 vs Pro: which is better?
In Google's human ratings 2.1 beats Pro on overall preference, infographics, editing and character consistency, at about a quarter of the price per 1K image. Pro still has more world knowledge, better brand consistency and accepts up to 3 style references.
How much does Nano Banana 2.1 cost?
On the Gemini API: $1.50 per million input tokens and $30 per million image-output tokens, which works out to about $0.034 for a 1K image, $0.050 for 2K and $0.113 for 4K. Batch is half price.
Can I use Nano Banana 2.1 on LazyKiwi?
Not at the moment. LazyKiwi's image generator offers Nano Banana alongside GPT Image, Seedream and Qwen Image.
Who makes Nano Banana?
Google. Nano Banana is the name of Gemini's native image models: Nano Banana (Gemini 2.5 Flash Image), Nano Banana Pro (Gemini 3 Pro Image), Nano Banana 2 (Gemini 3.1 Flash Image) and now Nano Banana 2.1. Platforms such as fal and OpenRouter resell access, but they do not make the model.
What is new in Nano Banana 2.1 compared with Nano Banana 2?
Google lists more accurate text and infographic layout, 1K/2K/4K output with fixes for wide and tall panoramas, and up to 14 reference images (up to 4 people and 10 objects kept consistent). Nano Banana 2's 0.5K size is gone, and per-image API prices are lower at every size.
Does Nano Banana 2.1 have a free tier?
Not on the Gemini API: Google lists no free tier for this model, and batch requests cost half the standard price. You can try Google's earlier Nano Banana model for free with new-account credits in LazyKiwi's image generator.
What resolution and aspect ratios does Nano Banana 2.1 support?
1K (default), 2K and 4K, in 14 aspect ratios from 1:8 to 8:1 — up to 4096 × 4096 at 1:1, according to Google's API documentation.
What does Nano Banana 2.1 still get wrong?
Google's model card notes that small text can blur, especially at 1K; long paragraphs render poorly; likeness is not always perfect; and left and right placement is sometimes swapped. Every image carries a SynthID watermark.
Try Google’s Nano Banana for free
Open LazyKiwi's image generator and try Nano Banana and other image models side by side.
More models
Image models you can run in LazyKiwi today.