Across 22 shared benchmarks, GPT-4o scores higher on 16 and LFM2.5-1.2B-Instruct on 4, with 2 level. The widest gap is AA Intelligence, where GPT-4o scores 11.2 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.