IntentGrasp - Policy Making: leaderboard

The Policy Making domain of the IntentGrasp (2026) All Set, intent-understanding items drawn from public intent datasets and posed as multiple choice with one or more correct intents. The score is instance F1 in percent, averaged over option-order shuffles; higher is better.

Metric: Instance F1 (%) on the Policy Making (PM) domain of the 12,909-item All Set, mean over option-order shuffles, zero-shot at temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around May 2028. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemma 4 31B (IT) [4bit]39.82#1803
2Llama 3.3 70B Instruct [4bit]34.52#1257
3Olmo-3.1-32B-Instruct [4bit]34.44
4gemma-4-E4B-it30.57#1539
5OLMo 3 7B Instruct26.75#6953
6Gemma 4 E2B (IT)26.69#2451
7Llama 3.1 8B Instruct25.26#4600
8Qwen 3 4B24.98#2348
9Qwen3-32B [4bit]24.93#1273
10Llama 3.2 3B Instruct21.22#7241
11Qwen 3 8B16.62#1873

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

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