What's new
By:

Introducing Nano Banana 2.1: Better Images at Half the Price

1000183389.webp
​

Nano Banana 2.1: Google's cheaper, sharper Flash image model​


Nano Banana 2.1 (gemini-nano-banana-2.1) is Google's new Flash-tier model for image generation and editing. It became generally available on October 6, 2026, and replaces Nano Banana 2 (gemini-3.1-flash-image), which shuts down on October 29.


What's new​


  • It's a high-efficiency model that keeps Flash-level speed and cost while improving visual quality, text rendering and multi-turn consistency.
  • Prompt adherence and character consistency across conversation turns are better than in Nano Banana 2.
  • It supports wide and panoramic aspect ratios (1:4, 4:1, 1:8, 8:1) at 1K, 2K and 4K, and tiling artifacts on these ratios at 2K and 4K are fixed.
  • It takes up to 14 reference images and keeps up to four characters and ten objects consistent.
  • It can ground images in Google Web Search and Google Image Search.
  • It has minimal, medium (default) and high thinking levels, which affect image quality.
  • Product recontextualization, mask- and ink-based editing, and factual accuracy are also improved.

Specs​


  • The input limit is 131,072 tokens and the output limit is 32,768 tokens. Inputs can be text, images, video and PDFs; outputs are images and text.
  • Video can serve as a reference, for example to generate a poster from a clip.
  • Output sizes are 1K (default), 2K and 4K. The 0.5K size Nano Banana 2 offered isn't supported.
  • It supports multi-turn editing, interleaved text and images, thinking, search grounding, C2PA Content Credentials and the Batch API. Every generated image carries a SynthID watermark.
  • It does not support seed, topK, logprobs, temperature or topP. Setting any of them returns an API error.
  • It also doesn't support function calling, structured outputs, code execution, file search, URL context, Maps grounding, the Live API, or Flex and Priority inference. If your pipeline needs those, let another Gemini model orchestrate and call this one for the image step.

Pricing and migration​


  • Prices are roughly halved versus Nano Banana 2. A 1K image drops from 6.70 to 3.36 cents, a 2K image from 10.10 to 5.04, and a 4K image from 15.10 to 7.56.
  • gemini-3.1-flash-image is deprecated and shuts down on October 29, 2026, so you'll need to migrate before then. 2.1 is the recommended model for all new projects.
  • It's rolling out across the Gemini app, AI Mode in Search, AI Studio, Flow, Stitch, Google Ads and the Gemini Enterprise platform.

How it compares​


Nano Banana Pro likely remains Google's best image model. In benchmarks, 2.1 with thinking scores above Pro on overall preference and infographic accuracy. In practice, though, Pro often still produces noticeably better images. In side-by-side tests, Pro looked more realistic, while 2.1 struggled with scale.


2.1 is the more efficient counterpart to Pro. For high-volume prompt testing, infographics and text-heavy designs, it looks like the cheaper default. For top-end quality, Pro may still win.


Published limits vary depending on where you look, with context windows ranging from 65,536 tokens to 1M, and input types differ too. Check the docs for the platform you're using.


Known limitations​


2.1 is built on Gemini 3.6 Flash and still has several weaknesses:


  • Small text is often blurry at 1K, and long paragraphs or page-length text render poorly.
  • Character consistency between input and output images isn't always perfect.
  • Masked and doodle-based edits can follow instructions only partly, and ink may persist.
  • Edits occasionally keep the input image's subject pose.
  • The model sometimes confuses spatial directions such as left and right.
  • World knowledge, 3D reasoning and factuality remain limited.

TRY NANO BANANA 2.1 NOW!​

You do not have permission to view the full content of this post. Log in or register now.

​

Read more about this release here:​

You do not have permission to view the full content of this post. Log in or register now.

​

Example:​

1000183433.webp
Nano Banana 2.1


1000183434.webp
Nano Banana Pro​


1000183077.webp
GPT Image 2.5​

Aerial from the control tower: Aerial wide shot from a drone-mounted camera above the launch complex at sunrise, a perfectly vertical plume column rising from the pad, radial shockwave across the scrubland, flocks of birds scattering, tiny technicians' vehicles for scale, shot on a DJI Inspire 3 with a 35mm full-frame lens, deep depth of field, soft golden haze, pin-sharp vehicle and pad detail, 8K, photoreal --ar 21:9 --style raw --v 7​



Your feedback is highly appreciated​

😎​

​

Support my other posts 🙏


 
Click to expand...

Similar threads

About this Thread

  • 2
    Replies
  • 35
    Views
  • 2
    Participants
Last reply from:
Diego Mendoza

New Topics

Trending Topics

Online now

Members online
1,098
Guests online
9,061
Total visitors
10,159

Forum statistics

Threads
2,336,790
Posts
29,293,693
Members
1,129,057
Latest member
makou
Back
Top