Claude Opus 5 Tops InferenceBench with 8.9x Speedup Over PyTorch

maksym_andr · x · 2026-08-13

InferenceBench has been updated with the latest frontier models, including Claude Opus 5, Sonnet 5, GPT-5.6 Sol Ultra, Kimi K2.7 Code, and Grok 4.5.

Claude Opus 5 claims the #1 spot, achieving an 8.90x geometric mean speedup over a naive PyTorch solution. Additionally, the cited tweet notes an interesting trend: models are now adapting their serving strategies dynamically based on the workload.

Original post →

More from Models

Models channel →