What is new in Nano Banana 2.1 compared with Nano Banana 2?

It adds multi-image fusion with up to 14 reference images, consistency for up to 4 characters and 10 objects, grounding through Google web and image search, and configurable thinking levels of minimal, medium, or high. It is an update to Nano Banana 2, which was released in May 2026.

What are the known limitations of Nano Banana 2.1?

Google lists weak rendering of small text, long paragraphs and full-page text, character consistency that is not always perfect between input and output images, occasional confusion over left-right spatial placement, and limited world knowledge, 3D reasoning and factual recall. Its Gemini 3.6 Flash base has a knowledge cutoff of March 2026, and in some domains knowledge may only go to January 2025.

Which models did Google compare Nano Banana 2.1 against?

The benchmarks compare Nano Banana 2.1 (Thinking) and Nano Banana 2.1 (No Thinking) with Gemini 3.1 Flash Image (Nano Banana 2 Thinking) and Gemini 3 Pro Image (Nano Banana Pro). Google says Nano Banana 2.1 (Thinking) scored highest across general preference, infographic design, infographic accuracy, and every editing category.

Large Language ModelsImage GenerationGIGAZINE AI4m agoOct 7, 2026, 13:00 JST

Google ships Nano Banana 2.1 image model with 14-image fusion

Send on LINE

3 Key Points

What happened
Google released Nano Banana 2.1, built on Gemini 3.6 Flash, adding support for up to 14 reference images, consistency for up to 4 characters, grounding via Google web and image search, and minimal/medium/high thinking levels.
Why it matters
Google says Nano Banana 2.1 wins every benchmark it published against Gemini 3.1 Flash Image and Gemini 3 Pro Image, which is likely aimed at teams that need reliable image editing rather than one-off generation.
What to watch
Google lists known limits such as weak rendering of small text and long paragraphs, imperfect character consistency, and confusion over left-right spatial placement, so the test is whether real editing jobs hit those gaps.

WHO IT HITSMarketing, design, and content teams that produce infographics or edit product imagery at scale are the clearest audience, since multi-image fusion and character consistency are the features Google says improved. Media and publishing staff working with AI-generated visuals may also weigh the documented limits on small text and layout accuracy.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

FAQ

What is new in Nano Banana 2.1 compared with Nano Banana 2?: It adds multi-image fusion with up to 14 reference images, consistency for up to 4 characters and 10 objects, grounding through Google web and image search, and configurable thinking levels of minimal, medium, or high. It is an update to Nano Banana 2, which was released in May 2026.
What are the known limitations of Nano Banana 2.1?: Google lists weak rendering of small text, long paragraphs and full-page text, character consistency that is not always perfect between input and output images, occasional confusion over left-right spatial placement, and limited world knowledge, 3D reasoning and factual recall. Its Gemini 3.6 Flash base has a knowledge cutoff of March 2026, and in some domains knowledge may only go to January 2025.
Which models did Google compare Nano Banana 2.1 against?: The benchmarks compare Nano Banana 2.1 (Thinking) and Nano Banana 2.1 (No Thinking) with Gemini 3.1 Flash Image (Nano Banana 2 Thinking) and Gemini 3 Pro Image (Nano Banana Pro). Google says Nano Banana 2.1 (Thinking) scored highest across general preference, infographic design, infographic accuracy, and every editing category.

GIGAZINE AIRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.