Nano Banana 2.1 Benchmark

Nano Banana 2.1 is Google's hosted model gemini-nano-banana-2.1. The rankings below are an automated snapshot from Artificial Analysis: blind user votes, Elo ratings, and a 95% confidence interval. This page does not add a separate first-party speed or quality table.

Google's image-generation docs recommend Nano Banana 2.1 for new projects and describe 1K, 2K, and 4K output. Nano Banana 2 remains gemini-3.1-flash-image. Neither model runs offline. Find the rows named Nano Banana 2.1 in the text-to-image and image-editing tables and read the Elo and confidence interval next to the refresh date; the next successful sync can change both the order and the scores. See how Elo works, what 4K means in the API, or the local AI image generator guide if you needed an offline model instead.

🕐 Data refreshed: November 2, 2026 · → Deploy locally

Filter: All models Current only ○ Open weights

#	Δ	Creator	Model	Elo	95% CI
1	—	OOpenAI	GPT Image 2.5 Sunburst (max)	1,222	±10
2	—	OOpenAI	GPT Image 2.5 Flare (max)	1,207	±10
3	—	OOpenAI	GPT Image 2 (high)	1,186	±11
4	—	GGoogle	Nano Banana 2.1	1,176	±12
5	—	MAMicrosoft AI	MAI-Image-2.6	1,166	±12
6	—	SSpaceXAI	Grok Imagine Image 2.0	1,166	±12
7	—	GGoogle	Nano Banana 2 (Gemini 3.1 Flash Image)	1,140	±10
8	—	MMeta	Muse Image	1,127	±11
9	—	OOpenAI	GPT Image 1.5 (high)	1,123	±11
10	—	MAMicrosoft AI	MAI-Image-2.6-Flash	1,119	±12

The leaderboard always shows the verified Elo ranking and confidence interval. Samples, release dates, and pricing are displayed when Artificial Analysis returns those fields through the Pro media endpoint; unavailable metadata is omitted instead of showing an empty column. Open-weight deployment guidance is maintained separately in the local deployment guides →

Filter: All models Current only ○ Open weights

#	Δ	Creator	Model	Elo	95% CI
1	—	OOpenAI	GPT Image 2.5 Sunburst (max)	1,199	±11
2	—	MAMicrosoft AI	MAI-Image-2.7-Preview	1,193	±15
3	—	OOpenAI	GPT Image 2.5 Flare (max)	1,182	±11
4	—	MAMicrosoft AI	MAI-Image-2.6	1,151	±12
5	—	KKrea	Krea 3 (Ultra)	1,151	±14
6	—	GGoogle	Nano Banana 2.1	1,151	±13
7	—	MAMicrosoft AI	MAI-Image-2.6-Flash	1,144	±13
8	—	OOpenAI	GPT Image 2 (high)	1,136	±13
9	—	BSByteDance Seed	Seedream 5.0 Pro	1,131	±13
10	—	MMeta	Muse Image	1,129	±12

Want to run these models locally? See deployment guides →

Cloud APIs vs Open Weights

This snapshot reports the ranking fields returned by Artificial Analysis. It does not infer a model's license, local availability, privacy, or total operating cost when those fields are absent; use the linked model cards and deployment guides for those questions.