Across 14 shared benchmarks, Claude 3.5 Haiku scores higher on 0 and K2 Horizon 375B A23B on 14. The widest gap is HLE, where K2 Horizon 375B A23B scores 32 against 3.6.
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.