Skip to content

Benchmark

ORCA-bench (Private)

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

Models measured
5
Scores recorded
5
Sources
1
IQ weight
0.56

Score distribution

21.8%
Mean
22.5%
Median
5.9%
Std dev
13.3%
10th percentile
29.6%
90th percentile
13.3%
Min
13.3%29.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.56 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
1Claude Opus 4.7orca-bench-private· Harbor Hub Leaderboards29.6%
2GPT 5.5orca-bench-private· Harbor Hub Leaderboards26.5%
3Claude Sonnet 4.6orca-bench-private· Harbor Hub Leaderboards22.5%
4GLM 5orca-bench-private· Harbor Hub Leaderboards17.4%
5DeepSeek V4 Proorca-bench-private· Harbor Hub Leaderboards13.3%