SpreadsheetBench v2 - Visualization Accuracy: leaderboard
Metric: Task accuracy (%). Source: spreadsheetbench.github.io. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 62.5 |
| 2 | GPT-5.2 | 45.83 |
| 3 | Gemini 3.1 Pro (Preview) | 41.67 |
| 4 | Kimi K2.5 | 41.67 |
| 5 | GLM-5 | 37.5 |
| 6 | DeepSeek V3.2 | 33.33 |
| 7 | Qwen 3.5 397B A17B | 25 |
| 8 | MiniMax-M2.5 | 16.67 |
Interactive version: theaggregate.ai/benchmark?slug=spreadsheetbench-v2-visualization-accuracy · How It Works · Data refreshed daily, snapshot 2026-10-02.