IntentGrasp - Writing: leaderboard

The Writing 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 Writing (W) 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 February 2028. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemma 4 31B (IT) [4bit]33.44#1803
2Llama 3.3 70B Instruct [4bit]33.06#1257
3Olmo-3.1-32B-Instruct [4bit]32.23
4OLMo 3 7B Instruct27.57#6953
5gemma-4-E4B-it27.41#1539
6Qwen3-32B [4bit]26.59#1273
7Qwen 3 4B23.94#2348
8Gemma 4 E2B (IT)23.36#2451
9Llama 3.2 3B Instruct21.14#7241
10Qwen 3 8B18.71#1873
11Llama 3.1 8B Instruct11.41#4600

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

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