OPMD Controls - Minerva - Pass@16: leaderboard
OPMD Controls - Minerva - Pass@16. OPMD arXiv2611.13855v1 main Table1. MATH-500 and Minerva,16 sampled responses/question, pass@16 and mean accuracy avg@16 as reported percentages. Only exact Qwen3-1.7B-Base student initialization and Qwen3-4B-Base/Qwen3-14B-Base teacher controls. NTP/OPMD trained step/anchor-selection arms retained in the exact source grid pending affirmative evaluated checkpoint bindings; not aliased to their student or teacher. Metric: Pass@16 (%). Higher is better.
Metric: Pass@16 (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3-14B-Base | 51.5 | #1504 |
| 2 | Qwen3-4B-Base | 34.9 | #2515 |
| 3 | Qwen3-1.7B-Base | 30.5 | #4890 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=opmd-controls-minerva-pass-16 · How It Works · Data refreshed daily, snapshot 2026-11-13.