Across 14 shared benchmarks, Agnes 2.5 Pro Alpha scores higher on 4 and Quasar 438B (max, based on GLM-5.2) on 10. The widest gap is OmniScience Non-Hallucination, where Quasar 438B (max, based on GLM-5.2) scores 78.6 against 11.9.
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.