Text to Image · Agentic

Best agentic AI image generators: AI Image Arena ranked

Nov 17, 2026 · 63,479 votes · 30 models

GPT-Image-2.5 Sunburst by OpenAI leads with 1403 points, ahead of GPT-Image-2.6 Sunburst (Medium) (1389) and GPT-Image-2.5 Flare (1363).

Lab
License
RankModelScoreSpreadVotes
1
GPT-Image-2.5 SunburstOpenAI · Proprietary
1403 ±11
1-211,914
2
GPT-Image-2.6 Sunburst (Medium)OpenAI · Proprietary
1389 ±27
1-5878
3
GPT-Image-2.5 FlareOpenAI · Proprietary
1363 ±10
2-512,013
4
GPT-Image-2.6 Flare (Medium)OpenAI · Proprietary
1362 ±26
2-6847
5
MAI-Image-2.7Microsoft AI · Proprietary
1340 ±45
2-8260
6
GPT-Image-2 (Medium)OpenAI · Proprietary
1335 ±13
4-74,187
7
MAI-Image-2.6Microsoft AI · Proprietary
1318 ±13
5-84,238
8
Grok Imagine Image 2.0xAI · Proprietary
1290 ±18
6-91,579
9
1263 ±12
8-125,357
10
1255 ±12
9-145,421
11
1240 ±13
9-163,549
12
Seedream 5.0 ProByteDance · Proprietary
1238 ±11
10-165,349
13
GPT-Image-1.5 (High Fidelity)OpenAI · Proprietary
1238 ±21
9-161,185
14
Qwen-Image-2.1Alibaba · qwen-research
1230 ±16
10-162,110
15
Muse ImageMeta · Proprietary
1228 ±12
11-164,654
16
MiniMax H3-ImageMiniMax · minimax-h3-image-community-license-agreement
1198 ±32
11-20464
17
Krea 2 MediumKrea · Proprietary
1173 ±12
16-205,083
18
Grok Imagine ImagexAI · Proprietary
1170 ±11
16-246,706
19
Krea 2 TurboKrea · krea-2-community-license
1169 ±13
16-243,368
20
Seedream 4.5ByteDance · Proprietary
1165 ±11
16-245,406
21
Wan 2.6Alibaba · Proprietary
1150 ±11
18-286,085
22
Ideogram 4.0 (Quality)Ideogram · Ideogram Open Model
1149 ±12
18-285,165
23
Seedream 5.0 LiteByteDance · Proprietary
1148 ±12
18-285,022
24
Krea 2 LargeKrea · Proprietary
1147 ±14
18-283,212
25
1139 ±11
21-295,269
26
UNI-1.1Luma AI · Proprietary
1138 ±13
21-303,459
27
FLUX.2 [pro]Black Forest Labs · Proprietary
1137 ±14
21-302,935
28
Wan 2.5 PreviewAlibaba · Proprietary
1133 ±14
21-303,223
29
Recraft V4Recraft · Proprietary
1120 ±12
25-305,147
30
HiDream-O1-ImageHiDream · MIT
1112 ±14
26-302,873

Score: Bradley-Terry rating - the bar shows it on one scale for the whole board, the thin line is the 95% confidence interval (± half of it next to the number). Spread: every rank the model could hold given that interval - models whose lines overlap are statistically tied.

Data: LMArena (lmarena-ai/leaderboard-dataset), CC BY 4.0. Published Nov 17, 2026. Flashlook mirrors the rankings unchanged - about the data.

Confidence intervals on model strength

Dot = score, line = 95% interval. Hover a dot for the numbers.

GPT-Image-2.5 Sunburst
GPT-Image-2.6 Sunburst (Medium)
GPT-Image-2.5 Flare
GPT-Image-2.6 Flare (Medium)
MAI-Image-2.7
GPT-Image-2 (Medium)
MAI-Image-2.6
Grok Imagine Image 2.0
Grok Imagine Image 2.0 (20260801)
Gemini 3.1 Flash Image (nano-banana-2, Web Search)
Gemini 3 Pro Image (nano-banana-pro, 2K)
Seedream 5.0 Pro
GPT-Image-1.5 (High Fidelity)
Qwen-Image-2.1
Muse Image
MiniMax H3-Image
Krea 2 Medium
Grok Imagine Image
Krea 2 Turbo
Seedream 4.5
Wan 2.6
Ideogram 4.0 (Quality)
Seedream 5.0 Lite
Krea 2 Large
Gemini 2.5 Flash Image Preview (nano-banana)
UNI-1.1
FLUX.2 [pro]
Wan 2.5 Preview
Recraft V4
HiDream-O1-Image
105011001150120012501300135014001450