gpt-4.1 is cheaper than claude sonnet 4.6 until caching enters the chat
huggingface's internal benchmarks reveal that model routing fails when optimizing for price alone, as caching mechanisms made claude sonnet 4.6 effectively half the cost of gpt-4.1 on agent tasks.
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.