Google released Nano Banana 2.1, an update to its Nano Banana 2 image model that cuts the price per image and improves text rendering and multi-turn character consistency. It runs on Gemini 3.6 Flash and keeps the Flash-tier speed of Nano Banana 2.
According to Google’s Gemini API documentation, the model (ID gemini-nano-banana-2.1) also improves visual quality and prompt adherence across 1K, 2K, and 4K output. Nano Banana 2.1 sits between Nano Banana 2 (Gemini 3.1 Flash Image) and Nano Banana Pro (Gemini 3 Pro Image). Google describes it as the more efficient counterpart to Nano Banana Pro, which remains its premium choice for the most complex visual tasks.
Nano Banana 2.1 Beats Nano Banana Pro in Google’s Own Preference Tests
The model card compares Nano Banana 2.1 against Nano Banana 2 and Nano Banana Pro using side-by-side human evaluations scored as Elo ratings. With thinking turned on, Nano Banana 2.1 leads every category Google lists.
| Benchmark | Nano Banana 2.1 (thinking on) | Nano Banana 2 | Nano Banana Pro |
|---|---|---|---|
| Overall preference | 1050 | 990 | 935 |
| Infographic design | 1048 | 961 | 912 |
| Infographic factuality (score) | 0.521 | 0.179 | 0.265 |
| General editing | 1026 | 938 | 939 |
| Multi-character consistency | 1106 | 978 | 1011 |
| Product consistency | 1024 | 955 | 965 |
| Stylization | 1062 | 991 | 990 |
| Multi-reference editing | 1066 | 988 | 989 |
Thinking mode accounts for a large part of the gap. With thinking off, Nano Banana 2.1 drops to 1015 on overall preference, which is still above Nano Banana 2 at 990 and Nano Banana Pro at 935. Infographic factuality falls from 0.521 to 0.328 without it.
Developers can pick from three thinking levels (minimal, medium, and high), with medium as the default. Google’s documentation lists these updates for the model:
- Better visual quality and realism at 1K, 2K, and 4K, with 1K as the default output size
- Fixes for tiling artifacts on wide and panoramic aspect ratios such as 1:4, 4:1, 1:8, and 8:1 at 2K and 4K
- Improved text rendering and infographic layout accuracy
- Multi-image fusion with up to 14 reference images, covering up to 4 characters and up to 10 objects
- Grounding with Google Web and Image Search
Nano Banana 2.1 Cuts the Price Per Image in Half at 1K and 2K Resolutions
Google’s Gemini API pricing page lists Nano Banana 2.1 at $1.50 per million input tokens, $7.50 per million text and thinking output tokens, and $30 per million image output tokens. Per image, standard pricing works out as follows:
| Output size | Nano Banana 2.1 | Nano Banana 2 | Nano Banana Pro |
|---|---|---|---|
| 1K image | $0.0336 | $0.067 | $0.134 |
| 2K image | $0.0504 | $0.101 | $0.134 |
| 4K image | $0.113 | $0.151 | $0.24 |
A 1K image from Nano Banana 2.1 costs a quarter of what Nano Banana Pro charges for the same size. The 4K saving is smaller, at about 25 percent against Nano Banana 2.
Batch API pricing for Nano Banana 2.1 is $0.0168 per 1K image, $0.0252 per 2K image, and $0.0567 per 4K image. There is no free tier for the model in the API, so it needs a paid account. Search grounding includes 5,000 free requests per month, shared across Gemini 3.x models, then $14 per 1,000 requests.
Nano Banana 2.1 Is Available in Six Google Products and Through the Gemini API
The model card lists the following distribution channels for Nano Banana 2.1:
- Gemini app
- Google AI Studio
- Gemini API
- Google Search AI Mode
- Google Ads
- Google Flow
- Google Stitch
Google Cloud’s documentation lists the model as generally available on the Gemini Enterprise Agent Platform. Seed, topK, logprobs, temperature, and topP parameters are not supported there and return an API error if set. Which Google plan covers these tools depends on how you use them, and our look at Google AI plans with family sharing covers what each one includes.
The Gemini API version accepts text, image, video, and PDF input and returns image and text output. Function calling, structured outputs, code execution, context caching, URL context, and the Live API are not supported, though the Batch API is.
Google’s deprecations page names Nano Banana 2.1 as the recommended replacement for gemini-3.1-flash-image, the model ID behind Nano Banana 2. No shutdown date for Nano Banana 2 has been announced.
Meet Nano Banana 2.1, our latest image generation and editing model 🍌
This upgraded version outperforms our previous models across the board, with notable leaps in visual design, mask-based editing, and subject consistency to help you create more natural-looking images.
— Google (@Google) October 6, 2026
Nano Banana 2.1 Still Has Trouble With Small Text and Left-Right Placement
The model card lists several known limitations, and Google does not hide them:
- Small text is often blurry at 1K, and long paragraphs and page-length text render poorly
- Character consistency between input and output images is not always perfect
- Masked and doodle-based edits can follow instructions only partially, and ink can persist
- Edits occasionally keep the subject’s original pose
- The model sometimes confuses spatial placement, such as left and right
- World knowledge, 3D reasoning, and factuality remain limited
The base model, Gemini 3.6 Flash, has a March 2026 knowledge cutoff, and Google says knowledge in some domains stops at January 2025.
OpenAI has been working the same area, with ChatGPT Images 2.0 adding thinking and better text rendering, followed by ChatGPT Images 2.5 with faster generation and new tools. Meta also entered the race with its first AI image generator, Muse Image.
