Reusable Prompts with variables

[ArnaudC2Care]

Hello,

I’m having trouble figuring out whether it’s possible to use Reusable Prompts with the responsesAPI, and if so, how to properly do it.

I'm referring to the example from the OpenAI documentation:

const response = await client.responses.create({
    model: "gpt-4.1",
    prompt: {
        id: "pmpt_abc123",
        version: "2",
        variables: {
            customer_name: "Jane Doe",
            product: "40oz juice box"
        }
    }
});

I’d like to use a reusable prompt in this way, injecting variables dynamically.

Thanks!

[StephenHodgson]

First you'll need to create the prompt in the playground, then copy that id there.

Then you'll pass it in the CreateReponse constructor:

var conversation = new List<IResponseItem>
{
    new Message(Role.User, "{{customer_name}} purchased {{product}}. What are some other recommendations we can give this customer?"),
};
var request = new CreateResponseRequest(
    input: conversation,
    model: Model.GPT4_1_Nano,
    prompt: new Prompt("pmpt_abc123", new Dictionary<string, object>
    {
        { "customer_name", "Jane Doe" },
        { "product", "40oz juice box" }
    }));
var response = await OpenAIClient.ResponsesEndpoint.CreateModelResponseAsync(request);

[ArnaudC2Care]

Thanks for your quick reply, it was that simple! 😄