OWE-Bench - Software: leaderboard

OWE-Bench - Software — 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
1Claude Fable 5.2Max0.6219#1
2Claude Opus 5.5Max0.4966#5
3GPT-6 AstraMax0.4426#6
4GPT-6.1 SolMax0.4314#10
5Claude Sonnet 5.5Max0.4218#19
6Claude Fable 5.1Max0.3661#15
7Gemini 4 CarbonHigh0.3151#2
8Gemini 4 ArgonHigh0.2214#12
9DeepSeek V4.1 FlashMax0.1474#62
10GLM-5.3Max0.1464#68
11Muse WonderMax0.142#28
12Grok 4.7xHigh0.08427#47
13Beam 1.0Max0.03616#206
14InklingMax0.01423#281

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