Skip to content

Benchmark

OmniDocBench - Read Order

OCR benchmarks measure how well models read text in images and documents. Scores are percentages. Lower is better.

Models measured
25
Scores recorded
25
Sources
1
IQ weight
1.90

Score distribution

20.7%
Mean
13.6%
Median
16.5%
Std dev
8.1%
10th percentile
40.1%
90th percentile
7.3%
Min
7.3%81%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.90 for chat models, measured over 23 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard