benchgap
Calibration

ERQA → OmniDocBench 1.5

OmniDocBench 1.5 is estimated from ERQA with a linear curve fitted on 7 models measured on both: y = 0.1604·x + 0.7913, R² = 0.45, cross-validated error 2.5 pp. It is used for 2 estimates.

Estimated modelERQAOmniDocBench 1.5Source
Glimpse83.8%92.6%estimated ± 2.5 pp, low confidence
Claude Opus 4.651.6%87.4%estimated ± 2.5 pp, low confidence

Fitted on

ModelERQAOmniDocBench 1.5
Claude Opus 4.546.8%87.7%
Gemini 3 Pro70.5%88.5%
GPT-5.259.8%85.7%
Qwen3.5 397B67.5%90.8%
Qwen3.7 Plus69.8%91.4%
Qwen3.8-27B65.5%91.1%
Qwen3.8 Max77.8%92.1%