How stateless is reasoning.encrypted_content?

[wong2333]

I’m using the Responses API with store: false and include: [‘reasoning.encrypted_content’], I want to confirm `reasoning.encrypted_content` Is it possible to use between different OpenAI providers?
So, for example, if the first request calls the official OpenAI API, the second request calls OpenRouter, and the third request calls Azure (for load balancing purposes), will they all succeed?

[_j]

How portable is encrypted content?

At the end of this post, I’ve provided a Responses API “output” for someone to place as input. See if another organization’s encryption and chat can be reused and is portable. Nobody’s taken me up yet and tried:

How to use reasoning.encrypted_content with store=False (stateless)

Thanks for an interesting observation. In your error, I wonder what “following item” is supposed to be referring to, as an “rs_” ID is for an entire reasoning container, which can have multiple array items inside. “Following” is language that would indicate something that comes after that “type”:“reasoning” object externally, such as “type”:“message”… Referencing a response.done for “output”, a single ID for all: "output": [ { "id": "rs_045fb12aa0dfc2ea0069170733badc8197969…

Experiment: You can send that back in your own Responses input and see if it is accepted or errors-out. You might not know if it was dropped, as OpenAI says they only utilize passed reasoning around tools (but that is inconsistent with a good caching policy).