Best AI for portraits: AI Image Arena ranked
Nov 17, 2026 · 1,263,222 votes · 86 models
GPT-Image-2.5 Sunburst by OpenAI leads with 1462 points, ahead of GPT-Image-2.5 Flare (1442) and GPT-Image-2.6 Flare (Medium) (1440).
| Rank | Model | Score | Spread | Votes |
|---|---|---|---|---|
| 1 | GPT-Image-2.5 SunburstOpenAI · Proprietary | 1462 ±9 | 1-3 | 10,999 |
| 2 | GPT-Image-2.5 FlareOpenAI · Proprietary | 1442 ±9 | 2-6 | 10,643 |
| 3 | GPT-Image-2.6 Flare (Medium)OpenAI · Proprietary | 1440 ±22 | 1-6 | 1,011 |
| 4 | GPT-Image-2.6 Sunburst (Medium)OpenAI · Proprietary | 1439 ±20 | 1-6 | 1,090 |
| 5 | GPT-Image-2 (Medium)OpenAI · Proprietary | 1427 ±7 | 2-6 | 20,336 |
| 6 | MAI-Image-2.7Microsoft AI · Proprietary | 1404 ±30 | 2-6 | 416 |
| 7 | MAI-Image-2.6Microsoft AI · Proprietary | 1345 ±9 | 7-9 | 7,432 |
| 8 | Gemini Nano Banana 2.1Google · Proprietary | 1337 ±20 | 7-11 | 968 |
| 9 | Grok Imagine Image 2.0xAI · Proprietary | 1334 ±14 | 7-11 | 1,937 |
| 10 | Grok Imagine Image 2.0 (20260801)xAI · Proprietary | 1323 ±9 | 8-11 | 6,906 |
| 11 | Reve 2.1Reve · Proprietary | 1319 ±15 | 8-11 | 1,552 |
| 12 | Muse ImageMeta · Proprietary | 1288 ±7 | 12-15 | 11,249 |
| 13 | Seedream 5.0 ProByteDance · Proprietary | 1288 ±7 | 12-15 | 21,891 |
| 14 | Reve 2.0Reve · Proprietary | 1282 ±11 | 12-15 | 3,104 |
| 15 | Qwen-Image-3.0-ProAlibaba · Proprietary | 1282 ±13 | 12-16 | 2,237 |
| 16 | GPT-Image-1.5 (High Fidelity)OpenAI · Proprietary | 1259 ±6 | 16-21 | 32,652 |
| 17 | Qwen-Image-2.1Alibaba · qwen-research | 1258 ±11 | 15-21 | 3,323 |
| 18 | Gemini 3 Pro Image (nano-banana-pro, 2K)Google · Proprietary | 1255 ±5 | 16-21 | 40,732 |
| 19 | MAI-Image-2.5Microsoft AI · Proprietary | 1255 ±7 | 16-21 | 11,214 |
| 20 | Gemini 3.1 Flash Image (nano-banana-2, Web Search)Google · Proprietary | 1254 ±6 | 16-21 | 16,334 |
| 21 | Gemini 3.1 Flash-Lite Image (nano-banana-2-lite)Google · Proprietary | 1243 ±11 | 16-22 | 3,509 |
| 22 | Gemini 3 Pro Image Preview (nano-banana-pro)Google · Proprietary | 1230 ±8 | 21-24 | 20,604 |
| 23 | Qwen-Image-2.0-Pro (2026-06-22)Alibaba · Proprietary | 1215 ±12 | 22-25 | 2,455 |
| 24 | MAI-Image-2Microsoft AI · Proprietary | 1207 ±8 | 23-25 | 11,221 |
| 25 | MiniMax H3-ImageMiniMax · minimax-h3-image-community-license-agreement | 1200 ±24 | 22-31 | 574 |
| 26 | Ideogram 4.0 (Quality)Ideogram · Ideogram Open Model | 1192 ±6 | 25-29 | 16,826 |
| 27 | Grok Imagine ImagexAI · Proprietary | 1184 ±5 | 25-30 | 64,273 |
| 28 | Cosmos3-Super-Text2Image (Agentic)NVIDIA · OpenMDW-1.1 | 1184 ±12 | 25-33 | 2,446 |
| 29 | UNI-1.1Luma AI · Proprietary | 1174 ±7 | 26-37 | 12,197 |
| 30 | Recraft V4.1 Utility ProRecraft · Proprietary | 1171 ±25 | 25-48 | 497 |
| 31 | Grok Imagine Image ProxAI · Proprietary | 1170 ±7 | 27-37 | 21,678 |
| 32 | Seedream 4.5ByteDance · Proprietary | 1166 ±5 | 29-39 | 67,397 |
| 33 | Imagen 4 Ultra (generate-001)Google · Proprietary | 1164 ±7 | 29-39 | 110,776 |
| 34 | Gemini 2.5 Flash Image Preview (nano-banana)Google · Proprietary | 1163 ±4 | 29-39 | 239,309 |
| 35 | UNI-1.1 MaxLuma AI · Proprietary | 1162 ±11 | 28-42 | 3,037 |
| 36 | Cosmos3-Super-Text2ImageNVIDIA · OpenMDW-1.1 | 1162 ±10 | 28-42 | 3,716 |
| 37 | Seedream 5.0 LiteByteDance · Proprietary | 1157 ±5 | 31-42 | 31,197 |
| 38 | FLUX.2 [max]Black Forest Labs · Proprietary | 1156 ±6 | 31-42 | 27,249 |
| 39 | Krea 2 MediumKrea · Proprietary | 1149 ±7 | 34-49 | 14,986 |
| 40 | Reve v1.5Reve · Proprietary | 1148 ±8 | 34-50 | 7,540 |
| 41 | HunyuanImage 3.0Tencent · tencent-hunyuan-community | 1148 ±6 | 34-49 | 42,340 |
| 42 | Recraft V4.1 ProRecraft · Proprietary | 1144 ±24 | 29-58 | 590 |
| 43 | FLUX.2 [flex]Black Forest Labs · Proprietary | 1144 ±6 | 38-52 | 34,248 |
| 44 | FLUX.2 [pro]Black Forest Labs · Proprietary | 1143 ±5 | 38-51 | 47,767 |
| 45 | Imagen 4 (generate-001)Google · Proprietary | 1143 ±5 | 38-52 | 144,786 |
| 46 | Qwen-Image (2512)Alibaba · Apache 2.0 | 1140 ±6 | 38-53 | 20,609 |
| 47 | Z-Image-TurboAlibaba · Apache 2.0 | 1139 ±9 | 38-55 | 5,475 |
| 48 | Seedream 4.0 (2K)ByteDance · Proprietary | 1139 ±11 | 38-57 | 3,513 |
| 49 | Krea 2 TurboKrea · krea-2-community-license | 1138 ±7 | 39-55 | 12,589 |
| 50 | Seedream 4.0 (High Res, fal)ByteDance · Proprietary | 1134 ±6 | 42-56 | 41,766 |
| 51 | FLUX.2 [dev]Black Forest Labs · flux-non-commercial-license | 1132 ±6 | 43-58 | 16,610 |
| 52 | Recraft V4.1 FlashRecraft · Proprietary | 1128 ±14 | 41-60 | 1,791 |
| 53 | Wan 2.7 Image ProAlibaba · Proprietary | 1127 ±8 | 45-59 | 7,084 |
| 54 | Wan 2.6Alibaba · Proprietary | 1127 ±5 | 46-58 | 54,878 |
| 55 | Wan 2.7 ImageAlibaba · Proprietary | 1124 ±8 | 46-59 | 7,204 |
| 56 | Recraft V4Recraft · Proprietary | 1123 ±5 | 49-59 | 33,753 |
| 57 | Krea 2 LargeKrea · Proprietary | 1122 ±7 | 48-60 | 12,725 |
| 58 | Seedream 4.0 (fal)ByteDance · Proprietary | 1116 ±11 | 50-62 | 3,576 |
| 59 | GPT-Image-1OpenAI · Proprietary | 1114 ±6 | 53-62 | 65,442 |
| 60 | GPT-Image-1 miniOpenAI · Proprietary | 1110 ±6 | 56-62 | 35,106 |
| 61 | HiDream-O1-ImageHiDream · MIT | 1104 ±6 | 58-64 | 15,437 |
| 62 | Seedream 3.0ByteDance · Proprietary | 1103 ±9 | 58-66 | 10,120 |
| 63 | P-ImageUnknown lab · Proprietary | 1098 ±6 | 61-67 | 24,276 |
| 64 | MAI-Image-1Microsoft AI · Proprietary | 1092 ±7 | 61-67 | 19,752 |
| 65 | FLUX.2 [klein] 9BBlack Forest Labs · flux-non-commercial-license | 1091 ±6 | 62-67 | 33,482 |
| 66 | Wan 2.5 PreviewAlibaba · Proprietary | 1091 ±5 | 62-67 | 62,375 |
| 67 | FLUX.1 Kontext [max]Black Forest Labs · Proprietary | 1088 ±6 | 63-69 | 19,646 |
| 68 | Qwen-ImageAlibaba · Apache 2.0 | 1078 ±5 | 67-69 | 28,633 |
| 69 | Imagen 3 (generate-002)Google · Proprietary | 1076 ±6 | 67-70 | 88,391 |
| 70 | FLUX.1 Kontext [pro]Black Forest Labs · Proprietary | 1067 ±6 | 69-70 | 95,741 |
| 71 | FLUX.2 [klein] 4BBlack Forest Labs · Apache 2.0 | 1055 ±6 | 71-74 | 33,620 |
| 72 | Qwen-Image (Prompt Extend)Alibaba · Apache 2.0 | 1053 ±5 | 71-75 | 198,021 |
| 73 | Ideogram 3.0 (Quality)Ideogram · Proprietary | 1052 ±7 | 71-76 | 28,126 |
| 74 | FLUX1.1 [pro]Black Forest Labs · Proprietary | 1044 ±7 | 71-76 | 14,171 |
| 75 | Lucid OriginLeonardo AI · Proprietary | 1042 ±6 | 72-76 | 73,413 |
| 76 | Recraft V3Recraft · Proprietary | 1042 ±6 | 73-76 | 43,272 |
| 77 | Luma PhotonLuma AI · Proprietary | 1025 ±7 | 77-79 | 31,372 |
| 78 | Ideogram 2.0Ideogram · Proprietary | 1020 ±7 | 77-80 | 14,547 |
| 79 | Runway Gen-4Runway · Proprietary | 1018 ±7 | 77-80 | 12,777 |
| 80 | FLUX.1 [dev] (FP8)Black Forest Labs · Open | 1000 ±8 | 80-81 | 10,614 |
| 81 | GLM-ImageZ.ai · MIT | 998 ±17 | 78-81 | 1,059 |
| 82 | Gemini 2.0 Flash Preview Image GenerationGoogle · Proprietary | 962 ±6 | 82 | 75,248 |
| 83 | FLUX.1 Kontext [dev]Black Forest Labs · flux-1-dev-non-commercial-license | 946 ±6 | 83-84 | 63,572 |
| 84 | Stable Diffusion 3.5 LargeUnknown lab · Open | 942 ±9 | 83-85 | 5,704 |
| 85 | BAGELByteDance · Apache 2.0 | 926 ±11 | 84-86 | 3,491 |
| 86 | DALL-E 3OpenAI · Proprietary | 926 ±7 | 85-86 | 60,218 |
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.
Portrait quality drives how believable an AI model looks in fashion photos. For a catalog, the same face also has to hold across every shoot - something one-off portraits don't test.
Shoot your products on-modelConfidence intervals on model strength
Dot = score, line = 95% interval. Hover a dot for the numbers.
Top 40 shown; the table lists all 86.

