Apertus-v1.1-0.5B: benchmark results

Provider: Apertus. Access: Open.

Aggregate Rating 1186 ± 34, rank #9497 of 10552 rated models, from 26 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Lara Mini Base Report (base-model protocol) - XNLI (English) (0-shot)54.1Accuracy (%), English benchmark65.7
Lara Mini Base Report (base-model protocol) - XNLI, All Languages (0-shot)40.5Accuracy (%) averaged over all languages/variants of the ben64.8
Lara Mini Base Report (base-model protocol) - Global PIQA (non-parallel), All Languages (0-shot)54.5Accuracy (%) averaged over all languages/variants of the ben39.8
Lara Mini Base Report (base-model protocol) - XCOPA, All Languages (0-shot)55.6Accuracy (%) averaged over all languages/variants of the ben36.1
Lara Mini Base Report (base-model protocol) - T-HellaSwag, All Languages (0-shot)34.1Accuracy normalized by answer length (%) averaged over all l16.7
Lara Mini Base Report (base-model protocol) - PIQA (0-shot)70.4Accuracy normalized by answer length (%), English benchmark7.4
Lara Mini Base Report (base-model protocol) - COPA (0-shot)74.8Accuracy (%), English benchmark6.5
Lara Mini Base Report (base-model protocol) - Global PIQA (parallel), All Languages (0-shot)22.6Accuracy (%) averaged over all languages/variants of the ben6.5
Lara Mini Base Report (base-model protocol) - Belebele (English) (0-shot)25.6Accuracy (%), English benchmark5.6
Lara Mini Base Report (base-model protocol) - CRUXEval-I (2-shot)18pass@1, direct prediction, official answer parser (%)5.6
Lara Mini Base Report (base-model protocol) - HellaSwag (10-shot)52.3Accuracy normalized by answer length (%), English benchmark3.7
Lara Mini Base Report (base-model protocol) - MATH (4-shot)3Math-Verify equivalence, Minerva prompts, 5,000 problems ove3.7

Interactive version: theaggregate.ai/model?slug=apertus-v1-1-0-5b · How It Works · Data refreshed daily, snapshot 2026-11-16.