Skip to content

Benchmark

LVOmniBench

Scores are percentages. Higher is better.

Models measured
3
Scores recorded
5
Sources
1
IQ weight
No weight published

Score distribution

66.3%
Mean
70.7%
Median
7.4%
Std dev
53.2%
10th percentile
73.6%
90th percentile
53.2%
Min
53.2%73.6%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Qwen3.8 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.73.6%
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.70.7%
3Gemini 3.8 FlashStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.70.7%
4Qwen3.8 Omni FlashStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.3%
5Qwen3.5 Omni PlusStatic· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.53.2%

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.