OfficeQA Pro v2 (release-reported)
Professional Work
OfficeQA Pro v2 is a revised version of the OfficeQA Pro split of Databricks' grounded-reasoning benchmark over U.S. Treasury Bulletin documents, first seen in Moonshot AI's Kimi K3.1 Pro launch post, where each task supplies the whole PDF corpus rendered as images with no machine-readable text. Its question count and the changes from OfficeQA Pro are not published, so it is a release-reported diagnostic separate from officeqa-pro-card.
Top models (higher is better)
| Model | Score |
|---|---|
| Kimi K3.1 Pro | 64.4 |