The Known Good Updated 25 Jul 2026

Image Arenas

Elo ratings from blind pairwise votes: a human is shown two anonymous outputs for the same prompt and picks the better one. The votes are collected and published by Arena — we ingest the ratings, we do not run the votes and we do not adjust them.

Text Image Video Speech

Image Arenas

Human preference rating from blind pairwise votes, ingested from Arena

Text to Image ArenaImage Editing Arena
Text to Image Arena

Elo from blind pairwise preference votes, collected and published by Arena. Higher is better. We ingest these ratings; we do not run the votes and we do not run our own evaluations.

12 of 79 models ▾ ☰⚙⇩ +Add model from specific provider
1385
1385
1302
1280
1271
1261
1261
1257
1250
1250
1245
1245
The Known Good
Text to Image Arena — full ranking
79 models rated 9,929,058 votes countedRatings captured 25 Jul 2026 Source: Arena
# Model Creator Elo 95% CI Votes Known Good Index
1 gpt-image-2 (medium) OpenAI 1,385 ±5 60,382 —
1 gpt-image-2 (medium) OpenAI 1,385 ±5 60,382 —
2 reve-2.1 reve 1,302 ±12 2,432 —
3 muse-image Meta 1,280 ±8 8,384 —
4 reve-2.0 reve 1,271 ±6 13,675 —
5 gemini-3.1-flash-image (nano-banana-2) [web-search] Google 1,261 ±7 18,502 —
5 gemini-3.1-flash-image (nano-banana-2) [web-search] Google 1,261 ±7 18,502 —
6 mai-image-2.5 microsoft-ai 1,257 ±5 34,416 —
7 gemini-3.1-flash-lite-image (nano-banana-2-lite) Google 1,250 ±8 9,392 —
7 gemini-3.1-flash-lite-image (nano-banana-2-lite) Google 1,250 ±8 9,392 —
8 gemini-3-pro-image-2k (nano-banana-pro) Google 1,245 ±3 129,467 —
8 gemini-3-pro-image-2k (nano-banana-pro) Google 1,245 ±3 129,467 —
9 gpt-image-1.5-high-fidelity OpenAI 1,240 ±3 133,306 —
10 gemini-3-pro-image-preview (nano-banana-pro) Google 1,232 ±5 82,565 —
10 gemini-3-pro-image-preview (nano-banana-pro) Google 1,232 ±5 82,565 —
11 seedream-5.0-pro ByteDance 1,231 ±11 2,865 —
12 grok-imagine-image-quality xAI 1,229 ±4 39,467 —
13 ideogram-4.0-quality Ideogram 1,207 ±6 18,547 —
14 qwen-image-2.0-pro-2026-06-22 Alibaba 1,193 ±7 8,419 —
15 uni-1.1-max Luma AI 1,188 ±6 13,458 —
16 mai-image-2 microsoft-ai 1,183 ±5 49,043 —
17 uni-1.1 Luma AI 1,177 ±5 20,320 —
18 grok-imagine-image xAI 1,173 ±3 205,635 —
19 recraft-v4.1-utility-pro Recraft 1,169 ±11 2,510 —
20 flux-2-max Black Forest Labs 1,162 ±4 117,111 —
21 grok-imagine-image-pro xAI 1,161 ±4 93,439 —
22 flux-2-flex Black Forest Labs 1,156 ±3 148,925 —
23 flux-2-pro Black Forest Labs 1,155 ±3 166,182 —
24 reve-v1.5 reve 1,154 ±4 31,694 —
25 gemini-2.5-flash-image-preview (nano-banana) Google 1,151 ±3 807,662 —
25 gemini-2.5-flash-image-preview (nano-banana) Google 1,151 ±3 807,662 —
26 hunyuan-image-3.0 Tencent 1,151 ±3 172,769 —
27 imagen-ultra-4.0-generate-001 Google 1,148 ±4 388,009 —
28 flux-2-dev Black Forest Labs 1,148 ±5 62,500 —
29 seedream-4.5 ByteDance 1,147 ±3 220,462 —
30 seedream-4-2k ByteDance 1,141 ±7 12,605 —
31 wan2.6-t2i Alibaba 1,134 ±3 171,024 —
32 seedream-5.0-lite ByteDance 1,132 ±4 87,149 —
33 recraft-v4.1-pro Recraft 1,130 ±11 2,674 —
34 imagen-4.0-generate-001 Google 1,129 ±3 531,357 —
35 qwen-image-2512 Alibaba 1,128 ±4 83,631 —
36 krea-2-medium krea 1,120 ±6 17,491 —
37 hidream-o1-image hidream 1,118 ±5 24,852 —
38 seedream-4-fal ByteDance 1,117 ±7 11,851 —
39 wan2.5-t2i-preview Alibaba 1,117 ±3 219,089 —
40 gpt-image-1 OpenAI 1,115 ±3 264,724 —
41 krea-2-turbo krea 1,113 ±7 9,143 —
42 seedream-4-high-res-fal ByteDance 1,113 ±3 174,974 —
43 recraft-v4 Recraft 1,113 ±4 89,987 —
44 gpt-image-1-mini OpenAI 1,109 ±3 163,465 —
45 wan2.7-image-pro Alibaba 1,102 ±5 28,673 —
46 krea-2-large krea 1,100 ±6 17,210 —
47 wan2.7-image Alibaba 1,099 ±5 29,016 —
48 mai-image-1 microsoft-ai 1,093 ±4 94,383 —
49 seedream-3 ByteDance 1,082 ±5 36,846 —
50 z-image-turbo Alibaba 1,082 ±6 19,864 —
51 flux-1-kontext-max Black Forest Labs 1,074 ±3 65,669 —
52 flux-2-klein-9b Black Forest Labs 1,069 ±3 142,277 —
53 qwen-image-prompt-extend Alibaba 1,060 ±3 704,537 —
54 flux-1-kontext-pro Black Forest Labs 1,059 ±3 330,789 —
55 imagen-3.0-generate-002 Google 1,058 ±3 360,193 —
56 Cosmos3-Super-Text2Image NVIDIA 1,057 ±6 18,247 —
58 qwen-image Alibaba 1,057 ±3 84,663 —
59 ideogram-v3-quality Ideogram 1,049 ±4 115,137 —
60 photon Luma AI 1,035 ±4 127,404 —
61 p-image Unknown 1,034 ±4 103,638 —
62 flux-2-klein-4b Black Forest Labs 1,030 ±3 144,073 —
63 runway-gen4 Runway 1,025 ±5 51,604 —
64 recraft-v3 Recraft 1,021 ±4 191,873 —
65 flux-1.1-pro Black Forest Labs 1,016 ±3 70,460 —
66 lucid-origin leonardo-ai 1,013 ±3 285,498 —
67 ideogram-v2 Ideogram 1,013 ±4 72,090 —
68 glm-image Z.ai 1,011 ±9 4,612 —
69 gemini-2.0-flash-preview-image-generation Google 975 ±3 257,095 —
70 flux-1-dev-fp8 Black Forest Labs 970 ±4 49,221 —
71 dall-e-3 OpenAI 968 ±4 239,288 —
72 flux-1-kontext-dev Black Forest Labs 940 ±4 215,355 —
73 stable-diffusion-v35-large Unknown 938 ±4 23,396 —
74 bagel ByteDance 898 ±6 12,423 —
The Known Good
Image Editing Arena — full ranking
58 models rated 61,824,267 votes countedRatings captured 25 Jul 2026 Source: Arena
# Model Creator Elo 95% CI Votes Known Good Index
1 gpt-image-2 (medium) OpenAI 1,465 ±4 145,970 —
1 gpt-image-2 (medium) OpenAI 1,465 ±4 145,970 —
2 muse-image Meta 1,402 ±6 16,932 —
3 mai-image-2.5 microsoft-ai 1,401 ±4 100,526 —
4 seedream-5.0-pro ByteDance 1,393 ±10 4,299 —
5 chatgpt-image-latest-high-fidelity (20251216) OpenAI 1,389 ±3 457,021 —
5 chatgpt-image-latest-high-fidelity (20251216) OpenAI 1,389 ±3 457,021 —
7 gemini-3-pro-image-2k (nano-banana-pro) Google 1,388 ±3 459,237 —
7 gemini-3-pro-image-2k (nano-banana-pro) Google 1,388 ±3 459,237 —
8 gemini-3-pro-image-preview (nano-banana-pro) Google 1,385 ±3 518,398 —
8 gemini-3-pro-image-preview (nano-banana-pro) Google 1,385 ±3 518,398 —
9 gemini-3.1-flash-image (nano-banana-2) [web-search] Google 1,385 ±4 60,941 —
9 gemini-3.1-flash-image (nano-banana-2) [web-search] Google 1,385 ±4 60,941 —
10 reve-2.1 reve 1,383 ±9 5,165 —
11 gpt-image-1.5-high-fidelity OpenAI 1,372 ±3 478,581 —
12 grok-imagine-image-quality xAI 1,359 ±4 111,066 —
13 reve-2.0 reve 1,358 ±7 14,513 —
14 uni-1.1-max Luma AI 1,334 ±5 40,859 —
15 grok-imagine-image xAI 1,329 ±3 478,004 —
16 qwen-image-2.0-pro-2026-06-22 Alibaba 1,315 ±5 32,739 —
17 gemini-3.1-flash-lite-image (nano-banana-2-lite) Google 1,314 ±5 26,611 —
17 gemini-3.1-flash-lite-image (nano-banana-2-lite) Google 1,314 ±5 26,611 —
18 uni-1.1 Luma AI 1,311 ±4 71,570 —
19 wan2.7-image-pro Alibaba 1,302 ±4 41,467 —
20 hunyuan-image-3.0-instruct Tencent 1,302 ±3 277,487 —
21 seedream-4.5 ByteDance 1,301 ±3 898,428 —
22 wan2.7-image Alibaba 1,301 ±4 42,318 —
23 gemini-2.5-flash-image-preview (nano-banana) Google 1,295 ±2 11,005,375 —
23 gemini-2.5-flash-image-preview (nano-banana) Google 1,295 ±2 11,005,375 —
24 seedream-5.0-lite ByteDance 1,294 ±3 271,130 —
25 seedream-4-2k ByteDance 1,271 ±7 213,066 —
26 flux-2-max Black Forest Labs 1,262 ±3 352,694 —
27 reve-v1.1 reve 1,261 ±3 631,952 —
28 kling-image-o1 Kuaishou 1,251 ±4 135,159 —
29 flux-2-pro Black Forest Labs 1,242 ±3 436,098 —
30 qwen-image-edit Alibaba 1,241 ±3 1,983,543 —
31 reve-v1 reve 1,234 ±5 380,892 —
32 qwen-image-edit-2511 Alibaba 1,233 ±3 401,599 —
33 wan2.6-image Alibaba 1,230 ±3 471,616 —
34 flux-2-flex Black Forest Labs 1,225 ±3 415,399 —
35 flux-2-klein-9b Black Forest Labs 1,224 ±3 506,759 —
36 flux-2-dev Black Forest Labs 1,220 ±4 188,220 —
37 seedream-4-high-res-fal ByteDance 1,217 ±3 1,184,210 —
38 p-image-edit Unknown 1,211 ±3 336,435 —
39 seedream-4-fal ByteDance 1,210 ±6 153,950 —
40 reve-v1.1-fast reve 1,206 ±3 521,652 —
41 reve-edit-fast reve 1,198 ±4 221,315 —
42 flux-2-klein-4b Black Forest Labs 1,188 ±3 506,764 —
43 flux-1-kontext-max Black Forest Labs 1,181 ±3 391,731 —
44 wan2.5-i2i-preview Alibaba 1,181 ±3 481,319 —
45 flux-1-kontext-pro Black Forest Labs 1,176 ±3 6,424,593 —
46 flux-1-kontext-dev Black Forest Labs 1,149 ±3 3,652,473 —
47 seededit-3.0 ByteDance 1,139 ±3 4,950,659 —
48 gpt-image-1 OpenAI 1,139 ±3 2,861,463 —
49 gpt-image-1-mini OpenAI 1,124 ±3 653,134 —
50 gemini-2.0-flash-preview-image-generation Google 1,081 ±2 4,966,089 —
51 bagel ByteDance 1,026 ±6 13,560 —
52 step1x-edit stepfun 998 ±4 155,733 —
The Known Good
How to read Elo. A rating is only meaningful against the population that produced it: a model's number in one arena cannot be compared to its number in another, and a rating built on a few hundred votes moves far more than one built on tens of thousands — the 95% confidence interval is the honest guide to that. Elo measures which output a person preferred, not whether it was correct; for correctness see the evaluations and the Known Good Index, which does not include Elo. Ratings are ingested from Arena and are published under the source's own terms — see attribution.