Benchmark Almanac

VideoScore-Bench v2

VideoScore2 44.4% tops its leaderboard Claude Sonnet 4 28.9% leads the models we track
Not in model cards
Version 2

Models score 500 generated videos on visual quality, agreement with the prompt, and physical consistency. The evaluation checks their ratings against human scores.

The EVA-RSI team ran this test separately from the original VideoScore-Bench. The public test file contains human ratings for all three dimensions.

Made by
TIGER Lab and collaborators
Released
September 2025
Kind
Video understanding
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Video → Text
Interaction
Single turn
Run by
Anyone
Based on
VideoScore-Bench
In views
Vision
Strongest model tested
GPT-5 #71 on the capability index
Last model added
August 2025
Scores refreshed
November 20th, 2026
16
Entries in its paper, 3 models we track

VideoScore2 tops the leaderboard with 44.4% Claude Sonnet 4 leads with 28.9% among models released by August 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. 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. The best model we track is Claude Sonnet 4, at 28.9%. This view shows only models we track, each at its best.

The newest model with a score here came out in August 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 VideoScore2 44.4%
  2. 2 VideoScore1.1 38.4%
  3. 3 VisionReward 36.6%
  4. 4 Dover 34%
  5. 5 DeQA-Score 32.5%
  6. 6 Q-Insight 31.7%
  7. 7 Claude Sonnet 4 28.9%
  8. 8 GLM-4.1V-9B-Thinking 28.7%
  9. 9 Q-Align 28.2%
  10. 10 ImageReward 28.1%
  11. 11 Gemini-2.5-Pro (snapshot unspecified) 27.9%
  12. 12 GPT-5 26.2%
  13. 13 VideoReward 26.2%
  14. 14 UnifiedReward 25.1%
  15. 15 VideoPhy2-AutoEval 22.4%
  16. 16 AIGVE-MACS 15.6%

Source: The VideoScore-Bench v2 paper

Source: The VideoScore-Bench v2 paper

#EntryScoreSettingDateSource
1 VideoScore2 44.4% Official leaderboard
2 VideoScore1.1 38.4% Official leaderboard
3 VisionReward 36.6% Official leaderboard
4 Dover 34% Official leaderboard
5 DeQA-Score 32.5% Official leaderboard
6 Q-Insight 31.7% Official leaderboard
7 Claude Sonnet 4 28.9% May 2025 Official leaderboard
8 GLM-4.1V-9B-Thinking 28.7% Jul 2025 Official leaderboard
9 Q-Align 28.2% Official leaderboard
10 ImageReward 28.1% Official leaderboard
11 Gemini-2.5-Pro (snapshot unspecified) 27.9% Official leaderboard
12 GPT-5 26.2% Aug 2025 Official leaderboard
13 VideoReward 26.2% Official leaderboard
14 UnifiedReward 25.1% Official leaderboard
15 VideoPhy2-AutoEval 22.4% Official leaderboard
16 AIGVE-MACS 15.6% Official leaderboard
#ModelScoreSettingDateSource
1 Claude Sonnet 4 28.9% May 2025 Official leaderboard
2 GLM-4.1V-9B-Thinking 28.7% Jul 2025 Official leaderboard
3 GPT-5 26.2% Aug 2025 Official leaderboard

Rankings by part

We read 3 parts from the benchmark’s paper: Visual Quality, Text Alignment and Physical/Common-sense Consistency. Each ranks models on one part of the test. GLM-4.1V-9B-Thinking leads 2 of them, and Claude Sonnet 4 leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

Visual Quality

3 models we track have a score here.

  1. 1 GLM-4.1V-9B-Thinking ↑1 33.3%
  2. 2 Claude Sonnet 4 ↓1 33.1%
  3. 3 GPT-5 30.7%

Number one over time

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

Anthropic
Benchmark paper Not tested since Sep 2025

Source: The VideoScore-Bench v2 paper

Scores over time

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

Plot with other benchmarks
In the benchmark’s paper
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Claude Sonnet 4 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Claude Sonnet 4

Source: Official VideoScore-Bench v2 leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

Versions

More in Audio and video

All 695 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%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for GPT-5. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (16 entries).
  • Pass Models scored 3 models we track, plus 13 other entries on its leaderboard.
  • Warning Newest score The newest came 15 months ago, in August 2025, for GPT-5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Pass Headroom The best is 28.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 3 rows for models we track and 13 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check