IntentGrasp - Gem Set - General: leaderboard

The General domain of the IntentGrasp (2026) Gem Set, hard intent-understanding items that every open-weight model failed in its first run. The score is instance F1 in percent, averaged over four option orders; higher is better.

Metric: Instance F1 (%) on the General (G) 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 February 2027. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)43.33#99
2Gemini 3 Flash (Preview)29.27#152
3Claude Opus 4.723.44#85
4Claude Sonnet 4.618.75#173
5GPT-5.4 Nano12.5#651
6GPT-5.410.42#135
7Claude Haiku 4.510.42#583
8Gemini 3.1 Flash Lite (Preview)8.85#417
9GPT-5.4 Mini6.25#406

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=intentgrasp-gem-set-general · How It Works · Data refreshed daily, snapshot 2026-11-21.