Responses API Code Interpreter

[mattbrandman]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

The responses API requires a "results" field on the code_interpreter model. This is populated with None on every response I have seen but is also required when sending the old code_interpreter blocks, however the api rejects the results field as unknown the class is

class ResponseCodeInterpreterToolCallParam(TypedDict, total=False):
    id: Required[str]
    """The unique ID of the code interpreter tool call."""

    code: Required[str]
    """The code to run."""

    results: Required[Iterable[Result]]
    """The results of the code interpreter tool call."""

    status: Required[Literal["in_progress", "interpreting", "completed"]]
    """The status of the code interpreter tool call."""

    type: Required[Literal["code_interpreter_call"]]
    """The type of the code interpreter tool call. Always `code_interpreter_call`."""

    container_id: str
    """The ID of the container used to run the code."""

To Reproduce

1. Make a request using code interpreter built in tool
2. Try to send a follow up request with the previous code interpreter block

Code snippets

OS

MacOS

Python version

Python v3.12

Library version

v1.84

[czgu]

+1 on this issue.

With the responses API, there is an "include" field that can be set to add outputs from the code_interpreter. It is documented here: https://platform.openai.com/docs/api-reference/responses/create#responses-create-include.

If this flag is enabled, a new "outputs" field will appear in the response, and its structure will be similar to the "results" field.

I also observed that in the OpenAI playground, when using the code_interpreter, the "outputs" field is always present, but there is no include flags in the generated CURL command to reproduce this behavior