Skip to content
IFAI

Model rankings

1512 models · 5 boards · 2 sources

Artificial AnalysisText-to-image arenaArenaText-to-image arena

Text-to-image arena · Anonymous head-to-head votes · Published September 24, 2026

Best images
gpt-image-2.5-sunburstOpenAI1,424Score
Best open model
qwen-image-2.1Alibaba1,228Score
Best from China
seedream-5.0-proByteDance1,256Score
Top three by Score
2gpt-image-2.5-flareOpenAI1,401
Top three open models
1qwen-image-2.1Alibaba · #17 overall1,228
2ideogram-4.0-qualityIdeogram · #18 overall1,205
3Cosmos3-Super-Text2Image (Agentic)NVIDIA · #25 overall1,165
This week's changes
claude-opus-5.5-maxCoding · #1New
Gemini 3.8 Flash TTSSpeech · #2New
Grok Imagine Image 2.0Image · #4New
gpt-6-sol-maxCoding · #5New
RankChangeModelScore95% intervalVotes
1—gpt-image-2.5-sunburstClosedOpenAI1,424±810,884
2—gpt-image-2.5-flareClosedOpenAI1,401±810,058
3—gpt-image-2 (medium)ClosedOpenAI1,383±488,744
4—mai-image-2.6ClosedMicrosoft AI1,335±618,625
5▲ 1reve-2.1ClosedReve1,302±88,305
6▼ 1grok-imagine-image-2.0 (low)ClosedxAI1,301±87,257
7—muse-imageClosedMeta1,276±535,422
8—reve-2.0ClosedReve1,269±615,802
9—gemini-3.1-flash-image (nano-banana-2) [web-search]ClosedGoogle1,261±453,228
10—seedream-5.0-proClosedByteDance1,256±485,690
11—qwen-image-3.0-proClosedAlibaba1,256±611,520
12—mai-image-2.5ClosedMicrosoft AI1,254±461,448
13—gemini-3.1-flash-lite-image (nano-banana-2-lite)ClosedGoogle1,250±617,851
14—gemini-3-pro-image-2k (nano-banana-pro)ClosedGoogle1,247±3168,447
15—gpt-image-1.5-high-fidelityClosedOpenAI1,238±3165,437
16—gemini-3-pro-image-preview (nano-banana-pro)ClosedGoogle1,234±586,451
17Newqwen-image-2.1OpenAlibaba1,228±94,445
18▼ 1ideogram-4.0-qualityOpenIdeogram1,205±453,371
19▼ 1qwen-image-2.0-pro-2026-06-22ClosedAlibaba1,192±613,123
20▼ 1uni-1.1-maxClosedLuma AI1,189±614,083
21▼ 1mai-image-2ClosedMicrosoft AI1,184±551,477
22▼ 1uni-1.1ClosedLuma AI1,183±441,412
23▼ 1grok-imagine-imageClosedxAI1,170±3262,132
24▼ 1recraft-v4.1-utility-proClosedRecraft1,169±112,618
25▼ 1Cosmos3-Super-Text2Image (Agentic)OpenNVIDIA · 1 more variant1,165±710,856
26—grok-imagine-image-proClosedxAI1,162±499,444
27▼ 2flux-2-maxClosedBlack Forest Labs1,162±3123,187
28▼ 1flux-2-flexClosedBlack Forest Labs1,157±3156,912
29—reve-v1.5ClosedReve1,155±439,755
30▼ 2flux-2-proClosedBlack Forest Labs1,154±3205,540
31—gemini-2.5-flash-image-preview (nano-banana)ClosedGoogle1,150±2878,451
32▼ 2hunyuan-image-3.0OpenTencent1,150±3180,550
33▼ 1imagen-ultra-4.0-generate-001ClosedGoogle1,148±4388,772
34▼ 1seedream-4.5ClosedByteDance1,147±3287,128
35▼ 1flux-2-devOpenBlack Forest Labs1,146±478,080
36▼ 1seedream-4-2kClosedByteDance1,140±712,566
37▼ 1seedream-5.0-liteClosedByteDance1,138±3123,178
38▼ 1wan2.6-t2iClosedAlibaba1,137±3223,535
39▼ 1recraft-v4.1-proClosedRecraft1,133±102,795
40▼ 1imagen-4.0-generate-001ClosedGoogle1,129±3539,788
41▼ 1qwen-image-2512OpenAlibaba · 1 more variant1,125±398,363
42▼ 1krea-2-mediumClosedKrea1,123±450,800
43▲ 1seedream-4-falClosedByteDance1,117±711,842
44▼ 2wan2.5-t2i-previewClosedAlibaba1,116±3271,922
45▼ 2hidream-o1-imageOpenHiDream1,116±459,511
46▼ 1gpt-image-1ClosedOpenAI1,116±3268,466
47▼ 1recraft-v4ClosedRecraft1,115±3129,674
48▼ 1seedream-4-high-res-falClosedByteDance1,113±3182,669
49▼ 1krea-2-turboOpenKrea1,113±543,913
50—krea-2-largeClosedKrea1,110±449,906

Intelligence over time

Each maker's best model at the time on the Artificial Analysis Intelligence Index, by the month it was released.

Gained most · Anthropic +57.2Highest now · Anthropic 64.6Gap between highest and lowest 18.1 ptsTime range Nov 2025 – Nov 2026
64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi

Arena data comes from the leaderboard dataset it publishes on Hugging Face (CC BY 4.0). Only each board's overall leaderboard is used; scores and ranks are unchanged. dataset · CC BY 4.0

Artificial Analysis data comes from its API (artificialanalysis.ai). Scores are unchanged; ranks follow the scores. artificialanalysis.ai

Open or closed follows the license Arena gives; prices and Artificial Analysis intelligence show only where model names match, otherwise “—”.