Bielik-1.5B-v3: benchmark results

Provider: SpeakLeash. Access: Open.

Aggregate Rating 1350 ± 26, rank #7715 of 10552 rated models, from 34 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Lara Mini Base Report (base-model protocol) - PAWS (English) (0-shot)66Accuracy (%), English benchmark68.5
Lara Mini Base Report (base-model protocol) - Belebele (English) (0-shot)66.2Accuracy (%), English benchmark52.8
Lara Mini Base Report (base-model protocol) - CRUXEval-I (2-shot)32.4pass@1, direct prediction, official answer parser (%)52.8
Lara Mini Base Report (base-model protocol) - MBPP (3-shot)41.8pass@1 (execution) (%)51.9
Lara Mini Base Report (base-model protocol) - MBPP+ (3-shot)47.6pass@1; must pass MBPP's own and the EvalPlus tests (%)50
Lara Mini Base Report (base-model protocol) - CRUXEval-O (2-shot)29.8pass@1, direct prediction, official answer parser (%)46.3
EuroEval Polish Knowledge30.41Knowledge Average Score (%)44.7
Lara Mini Base Report (base-model protocol) - MATH (4-shot)28.7Math-Verify equivalence, Minerva prompts, 5,000 problems ove42.6
EuroEval Polish NLU - PoQuAD43.35Character F1 (%)42.1
Lara Mini Base Report (base-model protocol) - MMLU (5-shot)55.2Accuracy (%), English benchmark40.7
EuroEval Polish NLU - Polemo279.16Matthews correlation coefficient (%)40.3
Lara Mini Base Report (base-model protocol) - Belebele, All Languages (0-shot)30.5Accuracy (%) averaged over all languages/variants of the ben38

Interactive version: theaggregate.ai/model?slug=bielik-1-5b-v3 · How It Works · Data refreshed daily, snapshot 2026-11-16.