OpenAI swap slashes LLM inference costs 40x with comparable quality

A platform engineer discovered that switching from GPT-4o to DeepSeek V4 Flash via Global API reduced LLM inference costs by 40x while maintaining comparable quality. The swap required only a two-line code change and met p99 latency budgets under 2.5 seconds. The team now saves more on inference than from three prior optimization sprints combined.