Research

Provider-price research notes and break-even models for AI workload economics, prompt caching, GPU utilization, and provider routing.