What changed: OpenAI added a 12-hour prompt caching option to the v1/responses API, keeping shared instructions, tools, and context cached longer so requests benefit from lower cached-input pricing even when hours apart. 12-hour cache writes cost 2.5x the input token price.
Who is affected
The source covers businesses using OpenAI's v1/responses API; no further restriction is stated.
What it means for your business
For businesses using the OpenAI API, this means they can keep repeated instructions and context cached longer to get lower cached-input pricing on requests spaced hours apart, at the cost of cache writes priced at 2.5x the input token rate.
What to do
Evaluate the 12-hour caching option for your API integration, weighing the 2.5x cache-write cost against your request pattern before choosing a cache duration.
When it takes effect
Available as of October 20, 2026.
Source: https://developers.openai.com/api/docs/changelog#2026-10-20
