OpenAI adds a 12-hour prompt caching option to its API

On 20 October 2026 OpenAI added a 12-hour lifetime for prompt caching to the Responses API, according to its API changelog. It applies to GPT-5.6 and later models. Developers set prompt_cache_options.ttl to 12h for a whole request, or set a ttl on an individual explicit cache breakpoint; leaving both out keeps the default of 30 minutes. OpenAI Developers pitched it at persistent agents, such as dots, that reuse the same instructions, tools and context throughout the day.

Pricing differs only on writes. A 30-minute cache write costs 1.25 times the uncached input rate and a 12-hour write costs 2.5 times; reads cost 0.1 times the input rate on most of these models and 0.05 times on GPT-6.1 Sol, whichever lifetime was chosen. The documentation says 12 hours is a minimum rather than a deletion deadline, that the option does not change data retention policy, and that a cache hit is not guaranteed for every request. Older models keep the separate prompt_cache_retention setting.

Shortly afterwards Tibo (@thsottiaux) of OpenAI wrote on X that logged-in users of Codex and Work now keep their cache indefinitely, so returning to an old conversation or resuming after a few hours carries no penalty and always hits the cache.