Skip to content

Benchmark

MMSU

Scores are percentages. Higher is better.

Models and systems measured
11
Scores recorded
12
Sources
2
IQ weight
0.67

Score distribution

66.4%
Mean
72.2%
Median
16%
Std dev
45%
10th percentile
83%
90th percentile
35.5%
Min
35.5%83.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.67 for chat models, measured over 9 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
1Gemini 3.8 FlashQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.3%
2Qwen3.5 Omni PlusQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83%
3Qwen3.5 Omni PlusQwen3.5-Omni release · published setting· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.8%
4Qwen3.8 Omni FlashQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.1%
5Qwen3.5 Omni FlashQwen3.5-Omni release · published setting· Reported by Qwen, Qwen Team· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.2%
6Gemini 1.5 Pro· MMSU data60.7%
7Qwen2.5-Omni 7B· MMSU data60.6%
8Muse Spark 1.2Qwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.59.9%
9Gemini 2.0 Flash· MMSU data51%
10Phi 4 Multimodal Instruct· MMSU data45%
11GLM-4-Voice· MMSU data35.5%

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-LiteQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.2%

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.