Best AI image editor for photorealistic edits: AI Image Arena ranked
Nov 12, 2026 · 3,429,805 votes · 49 models
GPT-Image-2.5 Sunburst by OpenAI leads with 1512 points, ahead of GPT-Image-2.5 Flare (1481) and MAI-Image-2.7-Preview (1468).
| Rank | Model | Score | Spread | Votes |
|---|---|---|---|---|
| 1 | GPT-Image-2.5 SunburstOpenAI · Proprietary | 1512 ±6 | 1 | 73,111 |
| 2 | GPT-Image-2.5 FlareOpenAI · Proprietary | 1481 ±6 | 2-4 | 71,154 |
| 3 | MAI-Image-2.7-PreviewMicrosoft AI · Proprietary | 1468 ±7 | 2-5 | 15,042 |
| 4 | GPT-Image-2.6 Sunburst (Medium)OpenAI · Proprietary | 1464 ±14 | 2-6 | 1,979 |
| 5 | GPT-Image-2.6 Flare (Medium)OpenAI · Proprietary | 1452 ±13 | 3-7 | 2,026 |
| 6 | GPT-Image-2 (Medium)OpenAI · Proprietary | 1450 ±4 | 4-6 | 186,269 |
| 7 | Grok Imagine Image 2.0xAI · Proprietary | 1432 ±7 | 6-8 | 20,077 |
| 8 | MAI-Image-2.6Microsoft AI · Proprietary | 1421 ±6 | 7-10 | 46,390 |
| 9 | Grok Imagine Image 2.0 (20260801)xAI · Proprietary | 1411 ±5 | 8-12 | 80,538 |
| 10 | Gemini Nano Banana 2.1Google · Proprietary | 1407 ±8 | 8-13 | 7,233 |
| 11 | MAI-Image-2.5Microsoft AI · Proprietary | 1403 ±5 | 9-13 | 110,670 |
| 12 | Muse ImageMeta · Proprietary | 1403 ±5 | 9-13 | 121,686 |
| 13 | Seedream 5.0 ProByteDance · Proprietary | 1392 ±4 | 13-16 | 225,822 |
| 14 | Grok Imagine Image (Quality, 20260519)xAI · Proprietary | 1391 ±6 | 13-17 | 23,300 |
| 15 | Krea 3 UltraKrea · Proprietary | 1390 ±14 | 10-18 | 1,636 |
| 16 | Gemini 3 Pro Image (nano-banana-pro, 2K)Google · Proprietary | 1386 ±4 | 13-17 | 201,733 |
| 17 | ChatGPT Image Latest (High Fidelity, 20251216)OpenAI · Proprietary | 1383 ±4 | 14-18 | 166,506 |
| 18 | Gemini 3.1 Flash Image (nano-banana-2, Web Search)Google · Proprietary | 1376 ±4 | 16-18 | 132,128 |
| 19 | GPT-Image-1.5 (High Fidelity)OpenAI · Proprietary | 1363 ±4 | 19-22 | 170,264 |
| 20 | Reve 2.1Reve · Proprietary | 1361 ±7 | 19-23 | 12,891 |
| 21 | Qwen-Image-2.1Alibaba · qwen-research | 1359 ±7 | 19-23 | 23,056 |
| 22 | MiniMax H3-ImageMiniMax · minimax-h3-image-community-license-agreement | 1351 ±11 | 19-23 | 3,142 |
| 23 | Reve 2.0Reve · Proprietary | 1351 ±7 | 20-23 | 12,291 |
| 24 | Ideogram 4.5Ideogram · Proprietary | 1334 ±5 | 24-25 | 67,599 |
| 25 | UNI-1.1 MaxLuma AI · Proprietary | 1325 ±6 | 24-26 | 26,385 |
| 26 | Grok Imagine ImagexAI · Proprietary | 1324 ±3 | 25-26 | 386,679 |
| 27 | UNI-1.1Luma AI · Proprietary | 1313 ±4 | 27-31 | 155,276 |
| 28 | HunyuanImage 3.0 InstructTencent · tencent-hunyuan-community | 1312 ±5 | 27-31 | 81,358 |
| 29 | Seedream 4.5ByteDance · Proprietary | 1306 ±3 | 27-31 | 423,837 |
| 30 | Gemini 3.1 Flash-Lite Image (nano-banana-2-lite)Google · Proprietary | 1304 ±6 | 27-31 | 39,684 |
| 31 | Qwen-Image-2.0-Pro (2026-06-22)Alibaba · Proprietary | 1303 ±6 | 27-31 | 35,732 |
| 32 | Seedream 5.0 LiteByteDance · Proprietary | 1291 ±3 | 32-33 | 236,168 |
| 33 | Gemini 2.5 Flash Image Preview (nano-banana)Google · Proprietary | 1290 ±3 | 32-33 | 403,806 |
| 34 | Reve v1.1Reve · Proprietary | 1269 ±5 | 34-35 | 57,225 |
| 35 | Kling Image O1Kling AI · Proprietary | 1259 ±5 | 34-37 | 37,871 |
| 36 | Qwen-Image-EditAlibaba · Apache 2.0 | 1256 ±5 | 35-38 | 37,160 |
| 37 | FLUX.2 [max]Black Forest Labs · Proprietary | 1255 ±5 | 35-38 | 37,263 |
| 38 | FLUX.2 [pro]Black Forest Labs · Proprietary | 1248 ±4 | 36-38 | 219,981 |
| 39 | Qwen-Image-Edit (2511)Alibaba · Apache 2.0 | 1233 ±4 | 39-41 | 114,522 |
| 40 | FLUX.2 [klein] 9BBlack Forest Labs · flux-non-commercial-license | 1230 ±5 | 39-41 | 60,290 |
| 41 | Wan 2.6 ImageAlibaba · Proprietary | 1229 ±4 | 39-41 | 221,191 |
| 42 | Reve v1.1 FastReve · Proprietary | 1216 ±5 | 42-46 | 46,533 |
| 43 | FLUX.2 [dev]Black Forest Labs · flux-non-commercial-license | 1214 ±6 | 42-46 | 49,655 |
| 44 | P-Image-EditUnknown lab · Proprietary | 1212 ±5 | 42-46 | 41,201 |
| 45 | Seedream 4.0 (High Res, fal)ByteDance · Proprietary | 1211 ±5 | 42-46 | 39,673 |
| 46 | FLUX.2 [flex]Black Forest Labs · Proprietary | 1208 ±5 | 42-46 | 39,900 |
| 47 | FLUX.2 [klein] 4BBlack Forest Labs · Apache 2.0 | 1189 ±5 | 47 | 60,555 |
| 48 | Wan 2.5 PreviewAlibaba · Proprietary | 1173 ±4 | 48 | 254,409 |
| 49 | GPT-Image-1 miniOpenAI · Proprietary | 1109 ±6 | 49 | 30,623 |
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 12, 2026. Flashlook mirrors the rankings unchanged - about the data.
Photorealistic edits from one source image - the core of turning flat-lays and mannequin shots into believable on-model photography.
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 49.

