granite-4.0-h-1B-base: benchmark results

Provider: IBM. Access: Open.

Aggregate Rating 1331 ± 38, rank #8065 of 10552 rated models, from 26 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Lara Mini Base Report (base-model protocol) - CRUXEval-O (2-shot)35.6pass@1, direct prediction, official answer parser (%)66.7
Lara Mini Base Report (base-model protocol) - CRUXEval-I (2-shot)34.3pass@1, direct prediction, official answer parser (%)63
Lara Mini Base Report (base-model protocol) - MATH (4-shot)32.5Math-Verify equivalence, Minerva prompts, 5,000 problems ove53.7
Lara Mini Base Report (base-model protocol) - MBPP+ (3-shot)47.9pass@1; must pass MBPP's own and the EvalPlus tests (%)51.9
Lara Mini Base Report (base-model protocol) - INCLUDE, All Languages (5-shot)40Accuracy (%) averaged over all languages/variants of the ben50
Lara Mini Base Report (base-model protocol) - Global MMLU, All Languages (5-shot)41.9Accuracy (%) averaged over all languages/variants of the ben48.1
Lara Mini Base Report (base-model protocol) - MBPP (3-shot)40.2pass@1 (execution) (%)48.1
Lara Mini Base Report (base-model protocol) - MGSM (8 (CoT)-shot)31.4Exact match on each language's native CoT answer phrase, ave47.2
Lara Mini Base Report (base-model protocol) - MMLU (5-shot)56.6Accuracy (%), English benchmark46.3
Lara Mini Base Report (base-model protocol) - PAWS (English) (0-shot)64.7Accuracy (%), English benchmark46.3
Lara Mini Base Report (base-model protocol) - HellaSwag (10-shot)69.7Accuracy normalized by answer length (%), English benchmark44.4
Lara Mini Base Report (base-model protocol) - PIQA (0-shot)76.3Accuracy normalized by answer length (%), English benchmark44.4

Interactive version: theaggregate.ai/model?slug=granite-4-0-h-1b-base · How It Works · Data refreshed daily, snapshot 2026-11-16.