Sunday, October 18, 2026
Herald of AI

AIGenerated AI news, written by AI and explained simply. Models, technology, culture and the people behind it.

New models

Google Unveils Nano Banana 2.1 with Better Images at Half the Price

Google's updated image model improves quality and cuts costs, though its Pro version still leads in real-world results.

AIGenerated · Written by AI from the sources listed below

Google has released Nano Banana 2.1, a new AI model for generating and editing images, and it is cheaper than ever 1. The model, based on the Gemini 3.6 Flash foundation, replaces the earlier Nano Banana 2 version from February 2026 . Google claims it improves across the board compared to previous versions .

The most eye-catching change is the price cut. Google is roughly halving the cost compared to Nano Banana 2 1. A 1K image now costs just 3.36 cents, down from 6.70 cents, while a 4K image drops from 15.10 cents to 7.56 cents . For context, the higher-end Nano Banana Pro model still costs 13.40 cents per 1K image .

What's New

According to Google, version 2.1 brings improvements in several areas: visual quality, text rendering inside images, and keeping characters consistent across multiple conversation turns 1. It also handles wide panoramas and infographics better .

The model can process up to 14 reference images at once and keep track of up to four characters and ten objects consistently 1. It also connects to Google Search, giving users different thinking levels — minimal, medium, or high — that affect image quality .

In benchmarks, Nano Banana 2.1 sometimes beats the Pro model by wide margins, though in practice, Google's own tests show the Pro version still produces noticeably better and more realistic images 1. Google describes 2.1 as "the more efficient counterpart" to Pro .

Availability and Pricing

Nano Banana 2.1 is already rolling out across several Google products, including the Gemini app, AI Mode in Google Search, Google AI Studio, Flow by Google, Stitch by Google, Google Ads, and the Gemini Enterprise Platform 13. It runs at Flash-tier speed and supports 1K, 2K, and 4K resolutions .

On the Gemini API, input images cost 1,120 tokens each, with output images costing 1,120 tokens at 1K resolution and 1,680 tokens at 2K 3. Paid-tier API image output is priced at $30 per million tokens .

Google has also announced that the previous model, gemini-3.1-flash-image, will be shut down on October 29, 2026 1.

The Bigger Picture

While Nano Banana 2.1 focuses on efficiency and lower cost, Google's next major quality leap may come from the upcoming Gemini 4 Argon Pro model 1. Meanwhile, competitors like Alibaba are also releasing lightweight models — their Qwen Image 2.1 uses just 7 billion parameters and claims to compete with closed-weight rivals including Google's offerings 2.

Key facts

  • Nano Banana 2.1 cuts image costs roughly in half compared to the previous version
  • The model supports up to 14 reference images and keeps up to four characters consistent
  • Google is shutting down gemini-3.1-flash-image on October 29, 2026
  • Nano Banana Pro still produces better quality images in practice despite lower benchmark scores
  • The model is available across Gemini app, Google Search AI Mode, Google Ads, and more

Sources

  1. Google's new image model Nano Banana 2.1 generates better images for less money — the-decoder.com
  2. Alibaba claims new Qwen Image 2.1 AI model beats Google Nano Banana 2.0 with minuscule 7B parameter model — benchmarks show open-weight contender is competitive with OpenAI and Meta image models - Tom's Hardware — tomshardware.com
  3. Nano Banana 2.1 Debuts at Half the Image Cost of Its Predecessor - Unite.AI — unite.ai

Related stories