OpenAI API Launches 12-Hour Prompt Cache
OpenAI announced that its API now supports 12-hour prompt caching, enabling persistent agents to reuse shared instructions, tools, and context throughout the day. Even with hours between requests, billing remains at the lower cached input rate. Codex and Work logged-in users enjoy permanent caching, allowing them to return to old conversations or resume after hours of suspension without penalty, with cache hits guaranteed.
OpenAI API Introduces 12-Hour Prompt Caching
OpenAI has announced that its API now supports 12-hour prompt caching. Persistent agents (like dots) can reuse shared instructions, tools, and context throughout the day, and even if requests are spaced several hours apart, they can still be billed at the lower cached input rate.
Additionally, logged-in users of Codex and Work enjoy permanent caching. This means you can return to old conversations or resume after a pause of several hours without any penalty, with a guaranteed cache hit.
OpenAI Developers(@OpenAIDevs) | Tibo(@thsottiaux)