Skip to content

Benchmark

CharXiv · RQ

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

Models measured
6
Scores recorded
6
Sources
1
IQ weight
0.59

Score distribution

80.3%
Mean
81%
Median
1.6%
Std dev
78.4%
10th percentile
82.5%
90th percentile
78%
Min
78%82.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.59 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1MiniMax M3.1 FlashMiniMax M3.1 Flash official-weights model-card run · reasoning effort, sampling, and run count not stated for this benchmark · reasoning questions · GPT-5.5 judge · code-interpreter setting not stated· Reported by MiniMax· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.5%
2Qwen3.6 PlusQwen official release table · thinking mode unless benchmark requires otherwise· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.81.5%
3MiMo V2.5Xiaomi MiMo-V2.5 official release/model card · RQ · Code Interpreter setting unspecified· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.81%
4MiMo V2 OmniXiaomi MiMo-V2-Omni release table · chart understanding · Code Interpreter setting unspecified· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.1%
5Qwen3.6 27BQwen official release table · thinking mode unless benchmark requires otherwise· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78.4%
6Qwen3.6 FlashQwen3.6-35B-A3B official model card · hosted API name Qwen3.6-Flash· Reported by Alibaba Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78%

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.