Mat-Pref - OOD-Property: leaderboard

Metric: Logical-question accuracy (%; majority vote over each logical question's distractor permutations, three- or four-choice ionic substitution questions graded against Materials Project DFT values; 510 band-gap logical questions on training-family hosts). Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.

Top models

#ModelScore
1Qwen 3 235B A22B 2507 Instruct53.3
2DeepSeek V350.6
3Llama 3.3 70B Instruct46.3
4Qwen 2.5 72B Instruct39.6
5Llama 3.1 8B Instruct27.8

Interactive version: theaggregate.ai/benchmark?slug=mat-pref-ood-property · How It Works · Data refreshed daily, snapshot 2026-09-26.