OpenAI Adds 12-Hour Prompt Cache Option for Models
The new cache setting allows developers to reuse shared context for GPT-5.6 models and later versions.
Updated on Oct. 21, 2026 in Artificial Intelligence
Live Poll
Do you plan to increase your use of AI tools despite rising costs for cached data?
OpenAI has introduced a 12-hour minimum prompt-cache lifetime for its GPT-5.6 and later models. The update enables developers to maintain shared context across requests made hours apart.
Why it matters
By allowing developers to match cache lifetimes to the frequency of context changes, the update optimizes costs and efficiency for complex API interactions. It also expands capabilities for power users by retaining context indefinitely for signed-in Codex and ChatGPT Work users.
The new 12-hour cache write costs 2.5 times the standard input-token rate, while the 30-minute option costs 1.25 times the input rate. Subsequent reads of cached context cost only one-tenth of the standard input rate.
The players
OpenAI
An artificial intelligence research organization known for developing advanced large language models like GPT.
The details
Developers can now configure their minimum cache lifetimes within the Responses API to suit their specific application needs. While users select either a 30-minute or 12-hour minimum, the system may retain cache entries for longer periods than those defined settings.
Timeline
October 20, 2026: OpenAI introduced the 12-hour minimum prompt-cache setting.
The Tech Race
This move represents a shift toward more flexible API architectures that prioritize cost efficiency for developers handling large-scale data. It positions OpenAI to better compete with cloud-native AI providers by reducing the overhead of repetitive prompt processing.
Developers using GPT-5.6 models can now improve workflow efficiency by minimizing redundant token costs for long-running context. However, users should carefully calculate the 2.5x write premium against their expected read frequency to ensure cost savings.
The takeaway
The addition of extended cache lifetimes provides a clearer roadmap for developers to optimize high-latency applications. It highlights a growing industry focus on balancing computational speed with the financial realities of token-based pricing.
Further reading
Find more insights on the Artificial Intelligence section page.
Source note: This article includes information reported by TokenPost.
Live Poll
Do you plan to increase your use of AI tools despite rising costs for cached data?
