gpt-4.1 比 claude sonnet 4.6 便宜，直到快取機制加入討論
huggingface 內部基準測試顯示，僅優化價格會導致模型路由失敗，因為快取機制使 claude sonnet 4.6 在代理任務中的有效成本僅為 gpt-4.1 的一半。
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.