GDP.xlsx: leaderboard
GDP.xlsx (Surge AI, 2026): 70 workplace questions, each paired with a professional spreadsheet (58 Excel and 12 CSV files in all, 393 worksheets, more than 2.1 million populated cells and fields) from finance, healthcare, engineering, real estate, the public sector and other domains. An agent inspects the file with Python spreadsheet libraries and LibreOffice in a sandboxed container (Harbor tasks, OpenHands-based harness, up to 200 turns and 60 minutes per task) and answers in its final message. Unlike SpreadsheetBench, which checks the cells an agent writes into a workbook, GDP.xlsx grades the written answer against task-specific must-have rubric criteria, and "It grades only the final response text: content that appears only in intermediate messages or in files the agent created is not credited." Score: mean rubric score (%) as Surge's evaluation harness README defines the benchmark score: a Gemini 3.8 Flash judge marks each criterion met or not, an attempt scores the share met, a task scores the mean of five attempts and the benchmark score is the unweighted mean over tasks (the README does not say outright that the leaderboard runs used these settings); reasoning effort as labelled. Higher is better. Source: https://surgehq.ai/benchmarks/gdp-xlsx
Metric: Mean rubric score (%). Source: surgehq.ai. Saturation forecast: Around 2029. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 38.3 | #9 (Gemini 4 Argon) |
| 2 | Claude Fable 5.1 (Max) | 23.1 | #10 (Claude Fable 5.1) |
| 3 | GPT-6 Astra (Max) | 22.9 | #6 (GPT-6 Astra) |
| 4 | Muse Spark 1.3 (Max) | 22.6 | #31 (Muse Spark 1.3) |
| 5 | Grok 4.7 (xHigh) | 22.3 | #26 (Grok 4.7) |
| 6 | GPT-6.1 Sol (Max) | 22 | #7 (GPT-6.1 Sol) |
| 7 | GLM-5.3 (Max) | 21.4 | #44 (GLM-5.3) |
| 8 | Qwen 3.8 Max (xHigh) | 20.3 | #34 (Qwen 3.8 Max) |
| 9 | Gemini 3.8 Flash (High) | 18.3 | #30 (Gemini 3.8 Flash) |
| 10 | Hy4 preview (High) | 18.3 | #45 (Hy4 preview) |
| 11 | GPT-6 Sol (Max) | 17.1 | #19 (GPT-6 Sol) |
| 12 | GPT-6 Luna (Max) | 14.9 | #61 (GPT-6 Luna) |
| 13 | Kimi K3 (Max) | 13.4 | #35 (Kimi K3) |
| 14 | Gemini 3.1 Pro (Preview) (High) | 10.3 | #64 (Gemini 3.1 Pro (Preview)) |
| 15 | DeepSeek V4 Pro (Max) | 6.9 | #112 (DeepSeek V4 Pro) |
Interactive version: theaggregate.ai/benchmark?slug=gdp-xlsx · How It Works · Data refreshed daily, snapshot 2026-11-05.