Sky-T1-32B-Preview: benchmark results

Provider: UC Berkeley. Released 2025-01-15. Access: Open.

Aggregate Rating 1596 ± 24, rank #1052 of 10916 rated models, from 155 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval German65.28Average Score (%)97.5
EuroEval English Common Sense Reasoning88.06Common Sense Reasoning Average Score (%)97.2
EuroEval Portuguese67.92Average Score (%)97
EuroEval Spanish62.67Average Score (%)96.4
EuroEval French70.64Average Score (%)96
EuroEval Italian66.96Average Score (%)96
EuroEval Portuguese NLU - MultiWikiQA PT78.11Character F1 (%)96
EuroEval English76.39Average Score (%)95.5
EuroEval German NLU61.07NLU Average Score (%)95.3
EuroEval Dutch73.32Average Score (%)95
EuroEval Italian Common Sense Reasoning77.8Common Sense Reasoning Average Score (%)94.7
EuroEval German NLU - Sb10K60.08Matthews correlation coefficient (%)94.2

Interactive version: theaggregate.ai/model?slug=sky-t1-32b-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.