IntentGrasp - Gem Set - Teaching: leaderboard

Metric: Instance F1 (%) on the Teaching (T) domain of the 470-item Gem Set, mean of four option-order runs, zero-shot, API default settings; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 9 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)43
2Gemini 3 Flash (Preview)39.67
3Claude Opus 4.733.5
4Gemini 3.1 Flash Lite (Preview)26.08
5Claude Haiku 4.512.5
6GPT-5.410.5
7Claude Sonnet 4.69.33
8GPT-5.4 Nano5.83
9GPT-5.4 Mini2

Interactive version: theaggregate.ai/benchmark?slug=intentgrasp-gem-set-teaching · How It Works · Data refreshed daily, snapshot 2026-10-07.