GPT-4.1はキャッシュが絡むまでClaude Sonnet 4.6より安い
Hugging Faceの社内ベンチマークによると、価格だけを最適化したモデルルーティングは失敗しうる。キャッシュ機構により、Claude Sonnet 4.6のエージェントタスクでの実質コストがGPT-4.1の半分になった。
huggingface tested model routing on 417 tasks using the appworld test challenge and found that claude sonnet 4.6 was nearly twice as cost-efficient as gpt-4.1 despite having higher base token prices and longer reasoning trajectories. the unexpected cost difference was driven by sonnet's aggressive cache-read pricing, which drastically lowered effective input costs during high-reuse agent workloads. the report concludes that effective routing systems must move beyond simple classification and optimize for the interaction between model pricing, workload characteristics, and infrastructure factors like latency and compliance.