캐싱이 등장하기 전까지는 GPT-4.1이 Claude Sonnet 4.6보다 저렴
Hugging Face의 내부 벤치마크에 따르면 가격만 최적화할 경우 모델 라우팅은 실패하며, 캐싱 메커니즘 덕분에 Claude Sonnet 4.6이 에이전트 작업에서 사실상 GPT-4.1 비용의 절반 수준이 되었습니다.
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.