Retrieve a response

You're Claude Code, working in my repository.

Task
Add a function to my app that fetches a stored response by id with GET /v1/responses/{response_id} and reports its status, output text and usage; for a background response, poll it until the status is completed, failed, cancelled or incomplete.

Read this first
Meta's docs page "Retrieve a response" as Markdown: https://dev.meta.ai/docs/api-reference/responses/retrieve-response.md
It's the source of truth for request and response shapes. The same page for people: https://dev.meta.ai/docs/api-reference/responses/retrieve-response

Setup
- Base URL: https://api.meta.ai/v1
- Auth: send Authorization: Bearer $MODEL_API_KEY, reading MODEL_API_KEY from the environment. Never hard-code, print or commit the key.
- Model: muse-spark-1.3-contributor (Contributor mode: Meta may use what you send to train future models; drop -contributor for Standard mode)
- Python: the OpenAI SDK (pip install openai) with base_url="https://api.meta.ai/v1" and api_key=os.environ["MODEL_API_KEY"] works, or plain requests.

Facts to get right
- Only stored responses (store: true, the default) can be retrieved.
- The reply is the Response with the request's parameters echoed (instructions, tools, metadata, max_output_tokens, previous_response_id, ...), incomplete_details, created_at, completed_at and the full usage.
- completed_at is set only once status is completed; it's null for queued, in_progress, failed and cancelled responses.
- Query parameters: stream=true (documented as SSE for in-progress background responses: response.created, a terminal event, [DONE]) and starting_after (skip events at or below a sequence_number). The live API answered stream=true with JSON when this was built, so handle both.
- Persistence can lag the terminal event: a GET right after completion may 404 for a few seconds, so retry once or twice with backoff.

Starter code (Python)
import os
import time

import openai
from openai import OpenAI

client = OpenAI(base_url="https://api.meta.ai/v1", api_key=os.environ["MODEL_API_KEY"])

def wait_for(response_id: str, tries: int = 90):
    for _ in range(tries):
        try:
            response = client.responses.retrieve(response_id)
            if response.status in ("completed", "failed", "cancelled", "incomplete"):
                return response
        except openai.NotFoundError:
            pass  # storage can lag a brand-new response for a moment
        time.sleep(2)
    raise TimeoutError(f"{response_id} is still running")

job = client.responses.create(model="muse-spark-1.3-contributor", input="Write a haiku about patience.", background=True)
response = wait_for(job.id)
print(response.status, response.completed_at, response.output_text, response.usage)

Done when
- A background response started by the script is polled until a terminal status, then its status, output text and usage print.
- The key comes from MODEL_API_KEY and never appears in code, logs or commits.
- API errors (4xx, 429) surface their message instead of failing silently.