INS-ActBench - Spreadsheet: leaderboard

Metric: Numerical accuracy (%; answer cells of the generated workbook within 0.2 percent relative tolerance) on the 636 spreadsheet tasks; 12,050 questions from the exams of 16 actuarial associations; knowledge and case questions two-shot, practice tasks zero-shot; proprietary models with thinking enabled at high where supported. Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.

Top models

#ModelScore
1GPT-5.5 (High)67.7
2Gemini 3.1 Pro (Preview)67
3Claude Opus 4.7 (High)60.6
4DeepSeek V4 Pro (High)55.2
5Qwen 3.6 Plus47.4
6Kimi K2.629.5
7Qwen 3.5 35B A3B16.3
8Qwen 3 14B2.7
9Gemma 3 12B (IT)1

Interactive version: theaggregate.ai/benchmark?slug=ins-actbench-spreadsheet · How It Works · Data refreshed daily, snapshot 2026-09-29.