Across 21 shared benchmarks, Granite 4.0 H 1B scores higher on 9 and LFM2.5-1.2B-Instruct on 9, with 3 level. The widest gap is scicode, where Granite 4.0 H 1B scores 8.2 against 2.3.
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.