Benchmark Almanac

LMArena Image-to-Video Arena

MiniMax H3 1,495 leads since Sep 2026 Not in model cards

Video generation models animate the same input image with the same prompt, anonymously and side by side. Votes give Elo-style ratings.

Shares votes infrastructure with the Text-to-Video Arena. Launched on the web on 2026-01-21.

Made by Arena (formerly LMArena) Released Jan 2026 Video generation All tasks public Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Ranked by votes People pick the better of two answers, as in an arena, and the votes are turned into ratings.
Answers
Open-ended Free-form text, documents, designs or media judged as a whole.
45
Entries on its own leaderboard, 44 models we track

MiniMax H3 leads with 1,495

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

Bars start at 900.

  1. 1 MiniMax H3 1,495
  2. 2 Gemini Omni 1.1 Flash 1,488
  3. 3 Wan 3.0 1,480
  4. 4 Dreamina Seedance 2.5 1,477
  5. 5 Dreamina Seedance 2.0 1,475
  6. 6 Gemini Omni Flash 1,465
  7. 7 HiDream-O1-Video 1.0 1,456
  8. 8 Grok Imagine Video 1.5 1,456
  9. 9 Flux 3 Video 1,449
  10. 10 HappyHorse 1.0 1,442
  11. 11 Wan 2.7 I2V 1,428
  12. 12 Grok Imagine Video 1,415
  13. 13 Veo 3.1 Audio 1,398
  14. 14 Veo 3.1 Fast Audio 1,385
  15. 15 Vidu Q3 Pro 1,363
  16. 16 Kling v3 Pro 1,354
  17. 17 Veo 3 Audio 1,331
  18. 18 Veo 3 Fast Audio 1,326
  19. 19 Wan 2.5 I2V Preview 1,321
  20. 20 Wan 2.6 I2V 1,310
  21. 21 Seedance v1.5 Pro 1,307
  22. 22 PixVerse v5.6 1,299
  23. 23 Kling 2.6 Pro 1,294
  24. 24 Kling 2.5 Turbo 1,276
  25. 25 Seedance v1 Pro 1,273
  26. 26 Hailuo 2.3 1,262
  27. 27 Veo 3 1,258
  28. 28 Veo 3 Fast 1,257
  29. 29 p-video 1,244
  30. 30 Vidu Q2 Turbo 1,244
  31. 31 Kling v2.1 Master 1,235
  32. 32 Hailuo 02 Pro 1,229
  33. 33 Kling v2.1 Standard 1,229
  34. 34 Ray 3 1,226
  35. 35 Hailuo 02 Standard 1,224
  36. 36 Vidu Q2 Pro 1,223
  37. 37 HunyuanVideo 1.5 1,198
  38. 38 Hailuo 02 Fast 1,194
  39. 39 Seedance v1 Lite 1,185
  40. 40 Wan 2.2 A14B 1,170
  41. 41 Veo 2 1,166
  42. 42 LTX-2 19B 1,159
  43. 43 Ray2 1,108
  44. 44 Runway Gen-4 Turbo 1,052
  45. 45 Pika v2.2 997
#EntryScoreSettingDateSource
1 MiniMax H3 1,495 Sep 2026 Official leaderboard
2 Gemini Omni 1.1 Flash 1,488 Sep 2026 Official leaderboard
3 Wan 3.0 1,480 Sep 2026 Official leaderboard
4 Dreamina Seedance 2.5 1,477 Sep 2026 Official leaderboard
5 Dreamina Seedance 2.0 1,475 Sep 2026 Official leaderboard
6 Gemini Omni Flash 1,465 Sep 2026 Official leaderboard
7 HiDream-O1-Video 1.0 1,456 Sep 2026 Official leaderboard
8 Grok Imagine Video 1.5 1,456 Sep 2026 Official leaderboard
9 Flux 3 Video 1,449 Sep 2026 Official leaderboard
10 HappyHorse 1.0 1,442 Sep 2026 Official leaderboard
11 Wan 2.7 I2V 1,428 Sep 2026 Official leaderboard
12 Grok Imagine Video 1,415 Sep 2026 Official leaderboard
13 Veo 3.1 Audio 1,398 Sep 2026 Official leaderboard
14 Veo 3.1 Fast Audio 1,385 Sep 2026 Official leaderboard
15 Vidu Q3 Pro 1,363 Sep 2026 Official leaderboard
16 Kling v3 Pro 1,354 Sep 2026 Official leaderboard
17 Veo 3 Audio 1,331 Sep 2026 Official leaderboard
18 Veo 3 Fast Audio 1,326 Sep 2026 Official leaderboard
19 Wan 2.5 I2V Preview 1,321 Sep 2026 Official leaderboard
20 Wan 2.6 I2V 1,310 Sep 2026 Official leaderboard
21 Seedance v1.5 Pro 1,307 Sep 2026 Official leaderboard
22 PixVerse v5.6 1,299 Sep 2026 Official leaderboard
23 Kling 2.6 Pro 1,294 Sep 2026 Official leaderboard
24 Kling 2.5 Turbo 1,276 Sep 2026 Official leaderboard
25 Seedance v1 Pro 1,273 Sep 2026 Official leaderboard
26 Hailuo 2.3 1,262 Sep 2026 Official leaderboard
27 Veo 3 1,258 Sep 2026 Official leaderboard
28 Veo 3 Fast 1,257 Sep 2026 Official leaderboard
29 p-video 1,244 Sep 2026 Official leaderboard
30 Vidu Q2 Turbo 1,244 Sep 2026 Official leaderboard
31 Kling v2.1 Master 1,235 Sep 2026 Official leaderboard
32 Hailuo 02 Pro 1,229 Sep 2026 Official leaderboard
33 Kling v2.1 Standard 1,229 Sep 2026 Official leaderboard
34 Ray 3 1,226 Sep 2026 Official leaderboard
35 Hailuo 02 Standard 1,224 Sep 2026 Official leaderboard
36 Vidu Q2 Pro 1,223 Sep 2026 Official leaderboard
37 HunyuanVideo 1.5 1,198 Sep 2026 Official leaderboard
38 Hailuo 02 Fast 1,194 Sep 2026 Official leaderboard
39 Seedance v1 Lite 1,185 Sep 2026 Official leaderboard
40 Wan 2.2 A14B 1,170 Sep 2026 Official leaderboard
41 Veo 2 1,166 Sep 2026 Official leaderboard
42 LTX-2 19B 1,159 Sep 2026 Official leaderboard
43 Ray2 1,108 Sep 2026 Official leaderboard
44 Runway Gen-4 Turbo 1,052 Sep 2026 Official leaderboard
45 Pika v2.2 997 Sep 2026 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

MiniMax
Own leaderboard 1 leader
MiniMax H3 1,495

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
1,000 1,200 1,400 1,000 1,200 1,400

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Audio and video

All 168 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%
  • TVBench In use Models answer multiple-choice questions on 10 temporal video tasks, such as action counting and ordering, that static frames cannot solve. Scored by accuracy. Qwen 3.8 Max 81.9%