For frontier AI news

Image Edit Arena

Compare models based on their ability to generate and edit images

Feb 7, 2026
23,202,840 votes
36 models
Rank Spread
1
11
1413±4
184,593
2
23
1400±4
179,642
3
24
1395±3
510,948
4
34
OpenAI · Proprietary
1390±4
202,523
5
56
xAI · Proprietary
1330±5
11,347
6
59
xAI · Proprietary
1322±7
7,183
7
69
Bytedance
Bytedance · Proprietary
1316±2
237,793
8
69
Tencent
Tencent · tencent-hunyuan-community
1315±4
50,075
9
69
1313±2
10,456,668
10
1010
Bytedance
Bytedance · Proprietary
1285±6
218,668
11
1111
Flux
Black Forest Labs · Proprietary
1267±3
109,294
12
1212
Reve
Reve · Proprietary
1261±2
227,779
13
1314
Flux
Black Forest Labs · Proprietary
1248±3
110,368
14
1316
Reve
Reve · Proprietary
1245±5
382,212
15
1416
Bytedance
Bytedance · Proprietary
1239±2
959,984
16
1416
Qwen Icon
Alibaba · Apache 2.0
1239±3
99,394
17
1719
Flux
Black Forest Labs · flux-non-commercial-license
1232±3
104,299
18
1719
Qwen Icon
Alibaba · Apache 2.0
1232±2
1,718,417
19
1719
Flux
Black Forest Labs · Proprietary
1231±3
85,555
20
2024
Qwen Icon
Alibaba · Proprietary
1222±4
48,422
21
2024
Flux
Black Forest Labs · Proprietary
1221±3
103,321
22
2024
Bytedance
Bytedance · Proprietary
1220±6
154,440
23
2024
Reve
Reve · Proprietary
1220±2
214,261
24
2024
Pruna AI
Pruna · Proprietary
1217±4
60,186
25
2525
Reve
Reve · Proprietary
1208±4
221,766
26
2628
Flux
Black Forest Labs · Apache 2.0
1194±3
104,522
27
2629
Qwen Icon
Alibaba · Proprietary
1191±4
78,611
28
2628
Flux
Black Forest Labs · Proprietary
1190±2
394,850
29
2829
Flux
Black Forest Labs · Proprietary
1185±2
6,475,424
30
3030
Flux
Black Forest Labs · flux-1-dev-non-commercial-license
1158±3
3,686,812
31
3132
OpenAI · Proprietary
1147±2
2,805,501
32
3132
Bytedance
Bytedance · Proprietary
1147±2
4,987,917
33
3333
OpenAI · Proprietary
1128±3
428,164
34
3434
1089±2
4,997,269
35
3535
Bytedance
Bytedance · Apache 2.0
1034±5
13,447
36
3636
Stepfun
StepFun · Apache 2.0
1006±4
156,077

Remove Style Control Leaderboard Plots

Average Win Rate Against All Other Models (Uniform Sampling and No Ties)

Fraction of Model A Wins for All Non-tied A vs. B Battles

Confidence Intervals on Model Strength (via Bootstrapping)

Battle Count for Each Combination of Models (without Ties)