Skip to content

Benchmark

ChartQAPro

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

Models measured
6
Scores recorded
8
Sources
2
IQ weight
2.10

Score distribution

68.7%
Mean
69.8%
Median
2.9%
Std dev
65.2%
10th percentile
70.9%
90th percentile
63.1%
Min
63.1%72.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.10 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
1Claude Opus 4.8Opus 4.8 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted · with tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.3%
2Doubao Seed 2.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.9%
3Doubao Seed 2.1 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.9%
4Claude Opus 4.7Opus 4.8 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted · with tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.69.8%
5Claude Opus 4.8Opus 4.8 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted · no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.69.4%
6Claude Opus 4.7Opus 4.8 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted · no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.67.6%
7Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim65.2%
8Mistral Large 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim63.1%

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. · Data by LLM Stats