PHYRD - Discover Closed Book - Astrophysics SAcc: leaderboard
PHYRD v1:63 Reproduce and43 Discover physics problems; complete problem-set denominators. SAcc averages each problem's correct-subquestion fraction, WAcc requires all subquestions correct. Missing/incomplete conclusions score0. Fields retain their native subsets. GPT-5.6 Sol uses a conversational interface with xhigh effort, distinct from four API models. Source-reported model identities retained without inventing dated endpoints. Discover Closed Book. API temperature0, requested high effort, initial65536 completion tokens escalated to131072 for missing final answers; Qwen thinking budget32768. GPT uses its conversational collection. No external retrieval/execution or reference solutions. Astrophysics; SAcc. Metric: Accuracy (%).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around January 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (xHigh) | 97.78 | #20 (GPT-5.6 Sol) |
| 2 | DeepSeek V4 Pro (High) | 82.78 | #109 (DeepSeek V4 Pro) |
| 3 | GLM-5.2 (High) | 61.11 | #78 (GLM-5.2) |
| 4 | Kimi K2.6 (High) | 55 | #112 (Kimi K2.6) |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=phyrd-discover-closed-book-astrophysics-sacc · How It Works · Data refreshed daily, snapshot 2026-10-27.