Responses API - No libraries (API only), how to submit the function result?

API
,
1

Note: I don’t use libraries, I use the API directly.

In the Responses API, when we create a model response with Functions, and the model responds with a function call like this:

{
  "id": "resp_67ca09c5efe0819096d0511c92b8c890096610f474011cc0",
  "object": "response",
  "created_at": 1741294021,
  "status": "completed",
  "error": null,
  "incomplete_details": null,
  "instructions": null,
  "max_output_tokens": null,
  "model": "gpt-4.1-2025-04-14",
  "output": [
    {
      "type": "function_call",
      "id": "fc_67ca09c6bedc8190a7abfec07b1a1332096610f474011cc0",
      "call_id": "call_unLAR8MvFNptuiZK6K6HCy5k",
      "name": "get_current_weather",
      "arguments": "{\"location\":\"Boston, MA\",\"unit\":\"celsius\"}",
      "status": "completed"
    }
  ],
 ...

How can I submit the function result? The documentation covers this, but while using Python or JavaScript libraries only, how can I do that using the API directly?

2

See if these prior writings satisfy your curiosity and need.

Note that if you get AI content and tool_call at the same time, they are different messages, completely breaking the paradigm of what actually happened.

You also must filter out reasoning summary items, or return them by ID and have had store=true and not had the storage expire on anything in the chat where you aren’t using the response id anyway and also have them not empty by not validating with personal ID and getting a lower class product for the same price. Or just plain no streaming on one model that your code can only obtain by failure.