[Enhancement]: Enhance Responses API support by keeping track of responses IDs and using them

[insilications]

What features would you like to see added?

First, thank you for implementing the Responses API. I'm not sure because it's been a while since I've worked with LibreChat, but it seems that to take full advantage of the Responses API we would need to return the id of the last Response object when making a request, for both the /api/agents/chat/:endpoint and /api/agents/chat/ flow:

import OpenAI from "openai";

const openai = new OpenAI();

const response = await openai.responses.create({
    model: "gpt-4o-mini",
    input: "tell me a joke",
    store: true,
});

console.log(response.output_text);

const secondResponse = await openai.responses.create({
    model: "gpt-4o-mini",
    previous_response_id: response.id,
    input: [{"role": "user", "content": "explain why this is funny."}],
    store: true,
});

console.log(secondResponse.output_text);

For the best results with this change, we recommend using the Responses API with the store parameter set to true, and passing in all reasoning items from previous requests (either using previous_response_id, or by taking all the output items from an older request and passing them in as input items for a new one). OpenAI will automatically include any relevant reasoning items in the model's context and ignore any irrelevant ones. In more advanced use‑cases where you’d like to manage what goes into the model's context more precisely, we recommend that you at least include all reasoning items between the latest function call and the previous user message. Doing this will ensure that the model doesn’t have to restart its reasoning when you respond to a function call, resulting in better function‑calling performance and lower overall token usage.
https://platform.openai.com/docs/guides/reasoning-best-practices#how-to-keep-costs-low-and-accuracy-high

PS: the default is that the store parameter is always true.

https://platform.openai.com/docs/guides/responses-vs-chat-completions
https://platform.openai.com/docs/guides/conversation-state

formatAgentMessages receives the incoming payload for a certain AgentClient and formats it as an array of messages for LangChain consumption.

LibreChat/api/server/controllers/agents/client.js

Line 706 in 24800bf

let { messages: initialMessages, indexTokenCountMap } = formatAgentMessages(

The key problem is that the incoming payload from the conversation's message history does not contain the Response id because it is obviously not saved in the database. To take full advantage of the Responses API, we would need to save it. LangChain makes things a bit easier: making formatAgentMessages return an AIMessage containing a response_metadata.id property seems to set previous_response_id when that AIMessage is used in subsequent ChatOpenAI Responses API calls:

ChatOpenAICallOptions:

  /**
   * The unique ID of the previous response to the model. Use this to create multi-turn
   * conversations. Will be set automatically if {@link ChatOpenAI.zdrEnabled} is `false`, provided
   * that AIMessages included in the request have a `response_metadata.id` property. Ignored unless
   * {@link ChatOpenAI.useResponsesApi} is `true`.
   *
   * If set, the Responses API will be used to fulfill the request.
   */
  previous_response_id?: ResponsesCreateParams["previous_response_id"];

https://github.com/langchain-ai/langchainjs/blob/0166a0411920d98ae610f6351a88dc17002f319d/libs/langchain-openai/src/chat_models.ts#L1202

invocationParams:
https://github.com/langchain-ai/langchainjs/blob/0166a0411920d98ae610f6351a88dc17002f319d/libs/langchain-openai/src/chat_models.ts#L2110

ChatOpenAI:

  /**
   * Must be set to `true` in tenancies with Zero Data Retention. Setting to `true` will disable
   * output storage in the Responses API, but this DOES NOT enable Zero Data Retention in your
   * OpenAI organization or project. This must be configured directly with OpenAI.
   *
   * See:
   * https://help.openai.com/en/articles/10503543-data-residency-for-the-openai-api
   * https://platform.openai.com/docs/api-reference/responses/create#responses-create-store
   *
   * @default false
   */
   zdrEnabled?: boolean | undefined;

https://github.com/langchain-ai/langchainjs/blob/0166a0411920d98ae610f6351a88dc17002f319d/libs/langchain-openai/src/chat_models.ts#L1933

More details

Another problem:

Data retention for model responses: Response objects are saved for 30 days by default.
https://platform.openai.com/docs/guides/conversation-state?api-mode=responses

We should handle situations where the Response id is older than 30 days and is therefore no longer valid. We can use the timestamp from the database to determine whether to pass all the messages and reasoning items from the history or the Response id.

Which components are impacted by your request?

Endpoints

Pictures

No response

Code of Conduct

• I agree to follow this project's Code of Conduct