{"slug": "gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks", "title": "GDP.xlsx: Benchmarking Frontier Models on Professional Workbooks", "summary": "A new benchmark, GDP_xlsx, finds that the best of 20 model configurations passes only 38.3% of realistic professional spreadsheet tasks, with a median score of 19.3%. The benchmark, described in arXiv paper 2611.17766v1, comprises 70 workplace requests written by professionals across 13 domains, paired with expert rubrics totaling 791 binary criteria and workbooks or CSV files holding 2.18 million populated cells; models are ranked by strict pass@1, the share of attempts meeting every criterion. The top configuration, Gemini 4 Argon, still leaves wide headroom, and the authors released the evaluation harness and three public tasks while holding out the other 67 to protect the leaderboard from contamination.", "body_md": "arXiv:2611.17766v1 Announce Type: new \nAbstract: 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.", "url": "https://wpnews.pro/news/gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks", "canonical_source": "https://www.machinebrief.com/news/gdpxlsx-benchmarking-frontier-models-on-professional-workboo-ptf1", "published_at": "2026-11-16 05:00:00+00:00", "updated_at": "2026-11-16 05:25:05.000465+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research"], "entities": ["GDP_xlsx", "Gemini 4 Argon", "Google", "Anthropic", "OpenAI", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks", "markdown": "https://wpnews.pro/news/gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks.md", "text": "https://wpnews.pro/news/gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks.txt", "jsonld": "https://wpnews.pro/news/gdp-xlsx-benchmarking-frontier-models-on-professional-workbooks.jsonld"}}