跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.17766 · v1

GDP.xlsx: Benchmarking Frontier Models on Professional Workbooks

首次提交 2026-11-13 13:24
版本更新 2026-11-13 13:24
雷达观测 2026-11-17 04:57
arXiv ↗PDF ↗

摘要

We introduce GDP_xlsx, a benchmark of 70 realistic workplace requests written by professionals in 13 domains, each paired with an expert rubric of binary criteria and with the complete workbook or CSV file it refers to. Together the files hold 2.18 million populated cells, and the rubrics 791 criteria. Agents work in a container with spreadsheet tools, and an agentic judge that can open the file applies the rubric. Completeness and exclusivity criteria require set-valued answers to be found in full and with nothing extra. We rank models by strict pass@1, the share of attempts that meet every criterion. Across 20 model configurations, including frontier models from Google, Anthropic, and OpenAI, the top-scoring one, Gemini 4 Argon, passes only 38.3% of attempts, and the median score is 19.3%, leaving wide headroom. We release the evaluation harness and three public tasks; the other 67 are held out to protect the leaderboard from contamination.

作者备注:14 pages, 4 figures, 6 tables. Accepted to the Agent Evaluation Science Symposium (New York City, November 2026). Leaderboard: https://surgehq.ai/benchmarks/gdp-xlsx ; evaluation harness: https://github.com/surge-ai/gdp-xlsx ; public tasks: https://huggingface.co/datasets/surgeai/GDP.xlsx

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v1 · 2026-11-13 13:24 · 最新观测版本

GDP.xlsx: Benchmarking Frontier Models on Professional Workbooks

We introduce GDP_xlsx, a benchmark of 70 realistic workplace requests written by professionals in 13 domains, each paired with an expert rubric of binary criteria and with the complete workbook or CSV file it refers to. Together the files hold 2.18 million populated cells, and the rubrics 791 criteria. Agents work in a container with spreadsheet tools, and an agentic judge that can open the file applies the rubric. Completeness and exclusivity criteria require set-valued answers to be found in full and with nothing extra. We rank models by strict pass@1, the share of attempts that meet every criterion. Across 20 model configurations, including frontier models from Google, Anthropic, and OpenAI, the top-scoring one, Gemini 4 Argon, passes only 38.3% of attempts, and the median score is 19.3%, leaving wide headroom. We release the evaluation harness and three public tasks; the other 67 are held out to protect the leaderboard from contamination.

查看此版本解读 →查看此版本原文 ↗