Benchmark Almanac

LMArena Text-to-Video Arena

Gemini Omni 1.2 Flash 1,540 leads since November 2026
Not in model cards

Video generation models make clips from the same text prompt, anonymously and side by side. Votes from logged-in users give Elo-style ratings.

Video Arena moved to the web on January 21st, 2026 with 15 video models. It ran earlier in another venue, per the blog.

Released
January 2026
Kind
Video generation
Level
Everyday
Scored
Ranked by votes
Answers
Open-ended
Tasks
All public
Last model added
November 2026
Scores refreshed
November 12th, 2026
48
Entries on its own leaderboard, 46 models we track

Gemini Omni 1.2 Flash leads with 1,540

Best score per entry from the benchmark's own leaderboard, read on November 12th, 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 1,000.

  1. 1 Gemini Omni 1.2 Flash 1,540
  2. 2 Gemini Omni 1.1 Flash 1,513
  3. 3 Gemini Omni Flash 1,509
  4. 4 Flux 3 Video 1,491
  5. 5 Grok Imagine Video 1.5 1,488
  6. 6 Dreamina Seedance 2.0 1,483
  7. 7 Dreamina Seedance 2.5 1,480
  8. 8 Wan 3.0 1,476
  9. 9 SkyReels V4 1,459
  10. 10 Muse Video 1,457
  11. 11 MiniMax H3 1,456
  12. 12 HappyHorse 1.0 1,429
  13. 13 Sora 2 Pro 1,367
  14. 14 Veo 3.1 Audio 1,363
  15. 15 Veo 3.1 Fast Audio 1,361
  16. 16 Veo 3 Fast Audio 1,347
  17. 17 Sora 2 1,342
  18. 18 Grok Imagine Video 1,341
  19. 19 Veo 3 Audio 1,340
  20. 20 Wan 2.7 T2V 1,337
  21. 21 Wan 2.6 T2V 1,324
  22. 22 Seedance v1.5 Pro 1,256
  23. 23 Veo 3 1,254
  24. 24 Veo 3 Fast 1,250
  25. 25 Wan 2.5 T2V Preview 1,244
  26. 26 PixVerse v5.6 1,238
  27. 27 Runway Gen-4.5 1,223
  28. 28 Kling 2.5 Turbo 1,219
  29. 29 Kling 2.6 Pro 1,216
  30. 30 P-Video 1,207
  31. 31 Ray 3 1,206
  32. 32 Hailuo 2.3 1,205
  33. 33 Kling O1 Pro 1,201
  34. 34 Hailuo 02 Pro 1,199
  35. 35 Seedance v1 Pro 1,192
  36. 36 Hailuo 02 Standard 1,181
  37. 37 Kandinsky 5.0 T2V Pro 1,175
  38. 38 HunyuanVideo 1.5 1,168
  39. 39 Veo 2 1,165
  40. 40 Kling v2.1 Master 1,163
  41. 41 LTX-2 19B 1,156
  42. 42 Wan 2.2 A14B 1,132
  43. 43 Seedance v1 Lite 1,113
  44. 44 Kandinsky 5.0 T2V Lite 1,113
  45. 45 Sora 1,069
  46. 46 Ray2 1,066
  47. 47 Pika v2.2 1,009
  48. 48 Mochi v1 1,007

Source: Official LMArena Text-to-Video Arena leaderboard

#EntryScoreSettingDateSource
1 Gemini Omni 1.2 Flash 1,540 Nov 2026 Official leaderboard
2 Gemini Omni 1.1 Flash 1,513 Nov 2026 Official leaderboard
3 Gemini Omni Flash 1,509 Nov 2026 Official leaderboard
4 Flux 3 Video 1,491 Nov 2026 Official leaderboard
5 Grok Imagine Video 1.5 1,488 Nov 2026 Official leaderboard
6 Dreamina Seedance 2.0 1,483 Nov 2026 Official leaderboard
7 Dreamina Seedance 2.5 1,480 Nov 2026 Official leaderboard
8 Wan 3.0 1,476 Nov 2026 Official leaderboard
9 SkyReels V4 1,459 Nov 2026 Official leaderboard
10 Muse Video 1,457 Nov 2026 Official leaderboard
11 MiniMax H3 1,456 Nov 2026 Official leaderboard
12 HappyHorse 1.0 1,429 Nov 2026 Official leaderboard
13 Sora 2 Pro 1,367 Nov 2026 Official leaderboard
14 Veo 3.1 Audio 1,363 Nov 2026 Official leaderboard
15 Veo 3.1 Fast Audio 1,361 Nov 2026 Official leaderboard
16 Veo 3 Fast Audio 1,347 Nov 2026 Official leaderboard
17 Sora 2 1,342 Nov 2026 Official leaderboard
18 Grok Imagine Video 1,341 Nov 2026 Official leaderboard
19 Veo 3 Audio 1,340 Nov 2026 Official leaderboard
20 Wan 2.7 T2V 1,337 Nov 2026 Official leaderboard
21 Wan 2.6 T2V 1,324 Nov 2026 Official leaderboard
22 Seedance v1.5 Pro 1,256 Nov 2026 Official leaderboard
23 Veo 3 1,254 Nov 2026 Official leaderboard
24 Veo 3 Fast 1,250 Nov 2026 Official leaderboard
25 Wan 2.5 T2V Preview 1,244 Nov 2026 Official leaderboard
26 PixVerse v5.6 1,238 Nov 2026 Official leaderboard
27 Runway Gen-4.5 1,223 Nov 2026 Official leaderboard
28 Kling 2.5 Turbo 1,219 Nov 2026 Official leaderboard
29 Kling 2.6 Pro 1,216 Nov 2026 Official leaderboard
30 P-Video 1,207 Nov 2026 Official leaderboard
31 Ray 3 1,206 Nov 2026 Official leaderboard
32 Hailuo 2.3 1,205 Nov 2026 Official leaderboard
33 Kling O1 Pro 1,201 Nov 2026 Official leaderboard
34 Hailuo 02 Pro 1,199 Nov 2026 Official leaderboard
35 Seedance v1 Pro 1,192 Nov 2026 Official leaderboard
36 Hailuo 02 Standard 1,181 Nov 2026 Official leaderboard
37 Kandinsky 5.0 T2V Pro 1,175 Nov 2026 Official leaderboard
38 HunyuanVideo 1.5 1,168 Nov 2026 Official leaderboard
39 Veo 2 1,165 Nov 2026 Official leaderboard
40 Kling v2.1 Master 1,163 Nov 2026 Official leaderboard
41 LTX-2 19B 1,156 Nov 2026 Official leaderboard
42 Wan 2.2 A14B 1,132 Nov 2026 Official leaderboard
43 Seedance v1 Lite 1,113 Nov 2026 Official leaderboard
44 Kandinsky 5.0 T2V Lite 1,113 Nov 2026 Official leaderboard
45 Sora 1,069 Nov 2026 Official leaderboard
46 Ray2 1,066 Nov 2026 Official leaderboard
47 Pika v2.2 1,009 Nov 2026 Official leaderboard
48 Mochi v1 1,007 Nov 2026 Official leaderboard

Number one over time

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

Google DeepMind
Own leaderboard 1 leader

Source: Official LMArena Text-to-Video Arena leaderboard

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 1,600 2027 1,000 1,200 1,400 1,600 2027

Source: Official LMArena Text-to-Video Arena leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Audio and video

All 233 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%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • 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%