OWE-Bench - Concrete Frame Sizing: leaderboard

OWE-Bench - Concrete Frame Sizing — Expert-built long-horizon science, software, ML and engineering tasks with task-specific hardware and4–18hour budgets; native baseline/ceiling normalized held-out reward,mean of8attempts and equal task means. Web frozen October1,2026;deterministic sealed verifier. Native model/effort and every attempt retained. Score fraction0–1;higher is better.

Metric: Mean normalized held-out task score (fraction, 0–1). Source: openworldengineer.mercor.com. Saturation forecast: Around August 2027. 14 models tracked. Updated 15 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-6 AstraMax0.5988#6
2GPT-6.1 SolMax0.5625#10
3Claude Fable 5.2Max0.5045#1
4Claude Opus 5.5Max0.3686#5
5Claude Sonnet 5.5Max0.334#19
6Gemini 4 CarbonHigh0.3162#2
7Claude Fable 5.1Max0.2843#15
8Muse WonderMax0.1639#28
9GLM-5.3Max0.1089#68
10DeepSeek V4.1 FlashMax0.08631#62
11Beam 1.0Max0.06539#206
12Grok 4.7xHigh0.05276#47
13Gemini 4 ArgonHigh0.04216#12
14InklingMax0.03849#281

Interactive version: theaggregate.ai/benchmark?slug=owe-bench-concrete-frame-sizing · How It Works · Data refreshed daily, snapshot 2026-11-16.