SpreadsheetBench v2: leaderboard
SpreadsheetBench 2: 321 expert-curated end-to-end business spreadsheet workflows spanning financial modeling, debugging and visualization, requiring multi-step coordinated operations and cross-sheet reasoning over complex multi-sheet workbooks. A separate and much harder task set than the original SpreadsheetBench board.
Metric: Score (%). Source: spreadsheetbench.github.io. Saturation forecast: Around August 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 34.89 | #63 |
| 2 | GPT-5.2 | 26.79 | #110 |
| 3 | Gemini 3.1 Pro (Preview) | 23.68 | #58 |
| 4 | GLM-5 | 17.14 | #141 |
| 5 | DeepSeek V3.2 | 15.58 | #205 |
| 6 | Kimi K2.5 | 14.64 | #144 |
| 7 | Qwen 3.5 397B A17B | 11.22 | #146 |
| 8 | MiniMax-M2.5 | 7.17 | #303 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=spreadsheetbench-v2 · How It Works · Data refreshed daily, snapshot 2026-10-20.