Datalab releases OmniParseBench, an open OCR benchmark with 16,288 pass/fail tests

Datalab releases OmniParseBench, an open OCR benchmark that scores document parsers with 16,288 yes/no tests on 2,937 pages across 92 languages.

The dataset and code are public, and the harness installs with uv pip install omni-parse-bench, needs Python 3.11+, and scores outputs offline without API keys. Gemini 3.8 Flash leads the launch leaderboard at 92.9, with Datalab accurate second at 92.7.