NOPO Study Baseline - Five Benchmark Average Pass16: leaderboard
The NOPO paper reports oracle pass@k for untouched, non-thinking Qwen3 baselines on coding, logic and mathematics tasks. The sample count k is part of each metric. The study’s weight-updated reinforcement-learning policies are separate artifacts and are not assigned to these original checkpoints. This board reports oracle pass@16 (%). Higher is better.
Metric: Oracle pass@16 (%). Source: arxiv.org. Saturation forecast: Around February 2027. 2 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Qwen 3 4B | Non-reasoning | 46.85 | #2343 |
| 2 | Qwen 3 8B | Non-reasoning | 44.94 | #1868 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=nopo-study-baseline-five-benchmark-average-pass16 · How It Works · Data refreshed daily, snapshot 2026-11-16.