OpenAI API supports 12-hour prompt caching; Tibo says Codex and Work cache will be retained permanently
BIBIBI
AT A GLANCE
PANews said OpenAI API supports 12-hour prompt caching, and Codex and Work cache for logged-in users can be retained permanently.
Article
OpenAI API launches 12-hour prompt caching,Codex and Work cache for logged-in users will be retained permanently
PANews October 21 reported, according to OpenAI Developers,OpenAI API now supports 12-hour prompt caching. Developers can cache shared instructions, tools, and context for longer periods and be charged at the lower cached-input rate even when requests are separated by several hours. This feature is primarily intended for agents that run continuously and reuse context throughout the day. Tibo (@thsottiaux) subsequently added that Codex and Work's logged-in users will have their cache retained permanently. Users can receive a cache hit when returning to an old conversation or resuming after several hours, with no additional charges.
PANews reported on October 21 that OpenAI API now supports 12-hour prompt caching.
02
The report, citing OpenAI Developers, said developers can cache shared instructions, tools, and context for longer periods.
03
The report stated that even when requests are separated by several hours, they can be charged at the lower cached-input rate.
04
The report said the feature is primarily intended for agents that run continuously and reuse context throughout the day.
05
Tibo（@thsottiaux) subsequently added that Codex and Work cache for logged-in users will be retained permanently.
06
The report stated that users can receive a cache hit when returning to an old conversation or resuming after several hours, with no additional charges.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
Prompt caching temporarily stores repeatedly used instructions, tools, and context. According to the report,OpenAI API's cache duration can reach 12 hours; Tibo also said that Codex and Work cache for logged-in users can be retained permanently.
Why it matters to readers
For agents that need to repeatedly use the same context, longer caching may reduce repeated input and allow eligible requests to be charged at the lower cached-input rate.
If you resume an old conversation after several hours or keep an agent running continuously, this change may affect billing for repeated context; however, the specific scope of applicability remains to be verified.
Risks and unknowns
The provided material does not include OpenAI's original announcement or pricing terms.
12 The applicable models, pricing, and effective date for -hour caching are not specified.
Codex and Work The specific scope of applicability of “permanently retained” and “no additional charges” is not specified.
The original source for Tibo's supplementary comments is not included in the provided material.
Related Developments
Loading event timeline…
Related concepts
Prompt Cache
This term is not in the glossary yet. Browse related concepts in the glossary.
