Better prompt caching for GPT-6

TL;DR - OpenAI says GPT-6 improves prompt caching to increase cache hit rates and reduce inference latency and costs. The announcement also highlights new diagnostics and controls for managing cache behavior.

• Higher cache hit rates should allow more repeated prompt content to reuse cached computation.
• New diagnostics provide greater visibility into prompt-caching behavior.
• Explicit breakpoints and additional controls give developers more influence over caching.
• The supplied description does not include benchmarks or implementation details.