SpreadsheetBench v2 - Debugging Cell Match: leaderboard
Metric: Cell-match score (%, Modif.) on the 100 debugging tasks (locating and repairing injected spreadsheet errors) under the SWE-agent-based scaffold (bash, view_xlsx and submit tools, at most 50 turns), thinking mode on, high reasoning effort for GPT-5.2 and Gemini 3.1 Pro. Source: arxiv.org. Saturation forecast: Around April 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Thinking) | 50.38 |
| 2 | Gemini 3.1 Pro (Preview) | 41.86 |
| 3 | GPT-5.2 (High) | 39.07 |
| 4 | GLM-5 | 29.06 |
| 5 | DeepSeek V3.2 (Thinking) | 28.71 |
| 6 | Qwen 3.5 397B A17B | 24.06 |
| 7 | Kimi K2.5 | 19.48 |
| 8 | MiniMax-M2.5 | 16.76 |
Interactive version: theaggregate.ai/benchmark?slug=spreadsheetbench-v2-debugging-cell-match · How It Works · Data refreshed daily, snapshot 2026-09-29.