Better prompt caching for GPT-6

OpenAI described improvements to GPT-6 prompt caching, including higher cache hit rates, new diagnostic tools, explicit breakpoints, and related control options, all aimed at reducing latency and cost.