Does Azure OpenAI support “Extended Prompt Cache Retention”?

Question

Does Azure OpenAI support “Extended Prompt Cache Retention”?

WayChou 0

Hello,

I would like to confirm whether Azure OpenAI Service supports the “Extended Prompt Cache Retention” feature described in the OpenAI documentation:

https://developers.openai.com/api/docs/guides/prompt-caching#extended-prompt-cache-retention

According to the OpenAI documentation, models such as gpt-5.2 support the prompt_cache_retention parameter for extended prompt cache retention.

However, when I attempt to use the same parameter with the gpt-5.2 deployment in Azure OpenAI, I receive the following error:

APIError: prompt_cache_retention is not supported on this model

x-request-id: 79f14e6e-ba6a-42ec-bb19-7d83cb6653f8

This raises a few questions:

Does Azure OpenAI currently support the Extended Prompt Cache Retention feature?
If not, is this feature planned for future support?
If yes, are there specific API versions, regions, or model deployment configurations required?

Any clarification would be greatly appreciated.

Thank you in advance for your support.

Nkudubeng Harry Mabula 0 Reputation points

2026-03-04T11:36:18.92+00:00

No — Azure OpenAI Service does not currently support** the “Extended Prompt Cache Retention” feature.** Azure supports only standard (in‑memory) prompt caching, which is cleared within 24 hours, and requests to use prompt_cache_retention on Azure models return an error.
Sridhar M 5,335 Reputation points Microsoft External Staff Moderator

2026-03-04T11:43:31.83+00:00
Hi WayChou

Azure OpenAI Service does not currently support the “Extended Prompt Cache Retention” feature (prompt_cache_retention) that is documented in the public OpenAI API documentation.

When the prompt_cache_retention parameter is sent to Azure OpenAI model deployments (including gpt-5.2), Azure returns an error such as:

prompt_cache_retention is not supported on this model

This behavior is expected and indicates that Azure OpenAI only supports standard (in‑memory) prompt caching, not extended retention

The OpenAI documentation you referenced applies to the OpenAI‑hosted API (platform.openai.com), not automatically to Azure OpenAI.

While OpenAI documents two cache retention modes:

In‑memory prompt caching

Extended prompt cache retention (configurable via prompt_cache_retention)

Azure OpenAI currently implements only the in‑memory variant, even for the same underlying model families (e.g., GPT‑5.x). The OpenAI documentation itself states that extended retention is available “when supported”, and Azure has not yet enabled that capability

Azure OpenAI does support prompt caching, but with the following characteristics:

Automatic, in‑memory caching only

Cache eligibility:

Prompt must be ≥ 1,024 tokens

First 1,024 tokens must be identical

Cache lifetime:

Cleared after inactivity

Always removed within 24 hours (Azure‑specific behavior)

No configuration required

No support for:

prompt_cache_retention

Extended / 24‑hour guaranteed retention policies

This is documented in Azure’s official prompt caching documentation. [learn.microsoft.com]

Azure OpenAI does not currently support Extended Prompt Cache Retention (prompt_cache_retention), even for GPT‑5.2. Only standard in‑memory prompt caching is available, and there is no documented timeline for extended retention support.

Thank you.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Sridhar M 5,335 Reputation points Microsoft External Staff Moderator

2026-03-05T12:19:43.05+00:00

Hi WayChou

Did you get any chance to review the above response.

Thank you!