Skip to content

Benchmark

VLMsAreBiased

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

Models measured
5
Scores recorded
5
Sources
2
IQ weight
1.60

Score distribution

72.1%
Mean
68.3%
Median
11.8%
Std dev
58.4%
10th percentile
88.3%
90th percentile
58.4%
Min
58.4%88.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.60 for chat models, measured over 5 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3.8 MaxQwen3.8-Max release vision/video table · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.88.3%
2Doubao Seed 2.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.6%
3Doubao Seed 2.1 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim68.3%
4Doubao Seed 1.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim62%
5Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim58.4%

Data credits: Data by LLM Stats · 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.