QMFOLBench: leaderboard

Metric: Macro-F1 (%; True, False and Unknown labels, 2,880 tasks). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview) (High)99.03
2GPT-5.4 (High)97.4
3Qwen 3.5 27B (Thinking)92.19
4DeepSeek V3.2 (Thinking)90.77
5Claude Sonnet 4.6 (Thinking)79.7
6DeepSeek V3.2 (Non-reasoning)76.27
7Qwen 3 32B (Thinking)63.29
8GPT-5.4 (Non-reasoning)55.08

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