Skip to content

Benchmark

Video-MME-v2

Vision benchmarks measure how well models understand images. Scores are percentages. Higher is better.

Models and systems measured
26
Scores recorded
89
Sources
3
IQ weight
1.91

Score distribution

28%
Mean
23.1%
Median
16.8%
Std dev
13.2%
10th percentile
55.9%
90th percentile
7.8%
Min
7.8%85.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.91 for chat models, measured over 25 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1MiniMax M3· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim85.4%
2Gemini 3.8 FlashAgent mode · Qwen Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.7%
3Qwen3.8 Omni FlashAgent mode · Qwen Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.71.3%
4Gemini 3.8 FlashStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.71%
5Qwen3.8 MaxQwen3.8-Max release vision/video table · subtitles enabled· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.68.3%
6Qwen3.8 Omni FlashStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.65%
7Muse Spark Open 816B A42B· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim61.5%
8Kimi K2.5· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim61.1%
9Qwen3.5 397B A17B· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim55.9%
10Gemini 3 Pro Previeww. sub @ 1fps· Video-MME-v249.4%
11Qwen3.5 Omni PlusStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.47.9%
12Doubao Seed 2.0 Prow. sub @ 1fps· Video-MME-v243.3%
13Gemini 3 Flashw. sub @ 1fps· Video-MME-v242.5%
14Kimi K2.5w. sub @ 512· Video-MME-v240.7%
15Qwen3.5 397B A17Bw. sub @ 512, think· Video-MME-v239.1%
16MiMo V2 Omniw. sub @ 1fps· Video-MME-v238.6%
17Gemini 3 Pro Previewwo sub @ 1fps· Video-MME-v238.2%
18GPT 5w. sub @ 50· Video-MME-v237%
19Kimi K2.5w. sub @ 64· Video-MME-v236.1%
20Doubao Seed 2.0 Prowo sub @ 1fps· Video-MME-v235.2%
21Kimi K2.5wo sub @ 512· Video-MME-v233%
22Gemini 3 Flashwo sub @ 1fps· Video-MME-v232.9%
23Qwen3.5 27Bw. sub @ 512, think· Video-MME-v231.4%
24Qwen3.5 397B A17Bw. sub @ 64, think· Video-MME-v230.6%
25Qwen3.5 397B A17Bwo sub @ 512, think· Video-MME-v230.3%
26MiMo V2 Omniwo sub @ 1fps· Video-MME-v229.9%
27Qwen3.5 27Bw. sub @ 64, think· Video-MME-v229.6%
28Gemma 4 31B ITw. sub @ 60· Video-MME-v228.2%
29Qwen3 VL 235Bw. sub @ 512, think· Video-MME-v228.1%
30Qwen3.5 Omni Plusw. sub @ 512· Video-MME-v227.4%
31Kimi K2.5wo sub @ 64· Video-MME-v227.4%
32Qwen3.5 122B-A10Bw. sub @ 64, think· Video-MME-v226.7%
33GPT 5wo sub @ 50· Video-MME-v226.4%
34Qwen3 VL 235Bw. sub @ 64, think· Video-MME-v226.3%
35Qwen3.5 9Bw. sub @ 512, think· Video-MME-v226.2%
36Qwen3.5 35B A3Bw. sub @ 64, think· Video-MME-v225.8%
37Qwen3.5 Omni Plusw. sub @ 64· Video-MME-v225.6%
38Qwen3.5 122B-A10Bw. sub @ 512, think· Video-MME-v225.1%
39Qwen3 VL 235Bw. sub @ 64, instruct· Video-MME-v225%
40Qwen3.5 397B A17Bw. sub @ 64, instruct· Video-MME-v224.5%
41Qwen3.5 27Bw. sub @ 64, instruct· Video-MME-v223.9%
42Qwen3.5 35B A3Bw. sub @ 512, think· Video-MME-v223.9%
43Qwen3.5 9Bw. sub @ 64, think· Video-MME-v223.2%
44internvl3-5-241b-a28bw. sub @ 64, instruct· Video-MME-v223.1%
45Qwen2.5 VL 72B Instructw. sub @ 64, instruct· Video-MME-v222.1%
46qwen3-vl-30bw. sub @ 512, think· Video-MME-v222.1%
47GLM 4.5Vw. sub @ 64, think· Video-MME-v221.8%
48qwen3-vl-30bw. sub @ 64, think· Video-MME-v221.1%
49Qwen3.5 122B-A10Bw. sub @ 64, instruct· Video-MME-v220.9%
50Qwen3.5 397B A17Bwo sub @ 64, think· Video-MME-v220.7%

Showing 1–50 of 88 results.

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Seed-2.0-LiteStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.64.9%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Hugging Face official benchmark leaderboards; licence varies by benchmark