OWE-Bench - Pairwise-Gated Attention Engine: leaderboard

OWE-Bench - Pairwise-Gated Attention Engine — 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 July 2027. 14 models tracked. Updated 15 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Fable 5.2Max0.6435#1
2Gemini 4 CarbonHigh0.5171#2
3Claude Sonnet 5.5Max0.4604#19
4Claude Opus 5.5Max0.4268#5
5GPT-6.1 SolMax0.3989#10
6Gemini 4 ArgonHigh0.3896#12
7Claude Fable 5.1Max0.368#15
8GPT-6 AstraMax0.3028#6
9DeepSeek V4.1 FlashMax0.2487#62
10GLM-5.3Max0.2374#68
11Muse WonderMax0.106#28
12Grok 4.7xHigh0.0457#47
13InklingMax0.0094#281
14Beam 1.0Max0.008125#206

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