SPU Public Task-adjusted Mean: leaderboard

SPU Public Task-adjusted Mean — Public100tasks/private16tasks reported separately. Mainmeans exclude safetyrefusals and unrecoverableoutputs; schema-recoverable rankings included unchanged. ScoresareweightedSpearmancorrelation over experimentgroups, bounds[-1,1]. Task-adjusted mean is fitted additive task/configuration estimate and not guaranteed bounded, so do not infer [-1,1] bounds for adjusted estimate. Native per-row taskcoverage retained. No metric from APIcost/table3. Metric: Spearman correlation. Source: https://arxiv.org/html/2611.15049v1

Metric: Spearman correlation. Source: arxiv.org. Saturation forecast: Around April 2028. 12 models tracked. Updated 16 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 5.5Max0.478#5
2GPT-6 AstraMedium0.431#6
3Claude Opus 5Max0.424#20
4Claude Fable 5.1Max0.421#15
5Claude Sonnet 5.5Max0.393#19
6Doubao-Seed-Evolving-0909Max0.315
7GPT-6.1 SolMax0.308#10
8Kimi K3.1 ProMax0.296#29
9DeepSeek V4.1 FlashMax0.254#62
10Hy4 previewMax0.245#71
11GLM-5.3 FlashMax0.239#125
12Qwen 3.8 Max (0902)Max0.227#49

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=spu-public-task-adjusted-mean · How It Works · Data refreshed daily, snapshot 2026-11-16.