NSMQ Riddles: leaderboard

Metric: Exact match (%; 156 riddles from the 2019 NSMQ riddles round, all clues given at once, answer checked against the answer and its listed variants). Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.

Top models

#ModelScore
1GPT-5.4 (High)86.54
2GPT-5.4 (Low)85.9
3Claude Opus 4.6 (High)84.62
4Gemini 3.1 Pro (Preview) (High)83.97
5GPT-OSS-120B (High)83.33
6Gemini 3.1 Pro (Preview) (Low)83.33
7GPT-OSS-120B (Low)80.77

Interactive version: theaggregate.ai/benchmark?slug=nsmq-riddles · How It Works · Data refreshed daily, snapshot 2026-09-25.