QMFOLBench - Food: leaderboard

Metric: Accuracy (%; 720 food-topic tasks). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview) (High)98.75
2GPT-5.4 (High)97.5
3DeepSeek V3.2 (Thinking)93.47
4Qwen 3.5 27B (Thinking)90.56
5DeepSeek V3.2 (Non-reasoning)78.75
6Claude Sonnet 4.6 (Thinking)78.61
7Qwen 3 32B (Thinking)63.89
8GPT-5.4 (Non-reasoning)57.5

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