Better prompt caching for GPT-6

• context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.

Higher cache hit rates and explicit breakpoints directly reduce latency and per-request cost for agents that repeatedly send large, mostly-unchanged context to GPT-6.

• postOpenAI Raises Default Prompt-Cache Hit Rates for GPT-6 AgentsOpenAIDevs
• postPrompt caching in LLMs, clearly explainedAvi Chawla
• videoGPT-5.6Bijan Bowen

Checking sign-in…

Loading comments…