S1.1 32B: benchmark results

Provider: Other.

Aggregate Rating 1519 ± 41, rank #2072 of 10916 rated models, from 98 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AbstentionBench - answer unknown - KUQ/Unsolved problems - Precision95.39Abstention Precision (%)100
AbstentionBench - false premise - FalseQA - Precision94.88Abstention Precision (%)100
AbstentionBench - underspecified context - BBQ - Precision99.63Abstention Precision (%)100
AbstentionBench - underspecified context - WorldSense - F194.57Abstention F1 (%)100
AbstentionBench - underspecified context - WorldSense - Precision90.66Abstention Precision (%)100
AbstentionBench - underspecified intent - KUQ/Ambiguous - Precision89.87Abstention Precision (%)100
AbstentionBench - answer unknown - KUQ/Future unknowns - Precision95.77Abstention Precision (%)94.7
AbstentionBench - false premise - KUQ/False assumptions - Precision96.12Abstention Precision (%)94.7
AbstentionBench - underspecified context - BB/Disambiguate - F165.69Abstention F1 (%)94.7
AbstentionBench - underspecified context - WorldSense - Recall98.84Abstention Recall (%)94.7
AbstentionBench - underspecified intent - SituatedQA/Geo - Precision78.43Abstention Precision (%)94.7
AbstentionBench - subjective - KUQ/Controversial - Precision95.52Abstention Precision (%)89.5

Interactive version: theaggregate.ai/model?slug=s1-1-32b · How It Works · Data refreshed daily, snapshot 2026-11-08.