Function request_o4mini

src/gpt_request.py:616–672

Makes a request to the o4-mini-2025-04-16 model with retry functionality. Args: prompt (str): The text prompt to send to the model log_id (str, optional): The log ID for tracking requests, defaults to tkb+timestamp max_tokens (int, optional): Maximum tokens for resp

(prompt, log_id=None, max_tokens=8000, max_retries=3, thinking=False)

def request_o4mini(prompt, log_id=None, max_tokens=8000, max_retries=3, thinking=False):
    """
    Makes a request to the o4-mini-2025-04-16 model with retry functionality.

    Args:
        prompt (str): The text prompt to send to the model
        log_id (str, optional): The log ID for tracking requests, defaults to tkb+timestamp
        max_tokens (int, optional): Maximum tokens for response, default 8000
        max_retries (int, optional): Maximum number of retry attempts, default 3
        thinking (bool, optional): Whether to enable thinking mode, default False

    Returns:
        dict: The model's response
    """
    base_url = cfg("gpt4omini", "base_url")
    api_version = cfg("gpt4omini", "api_version")
    ak = cfg("gpt4omini", "api_key")
    model_name = cfg("gpt4omini", "model")

    client = openai.AzureOpenAI(
        azure_endpoint=base_url,
        api_version=api_version,
        api_key=ak,
    )

    if log_id is None:
        log_id = generate_log_id()

    extra_headers = {"X-TT-LOGID": log_id}

    # Configure extra_body for thinking if enabled
    extra_body = None
    if thinking:
        extra_body = {"thinking": {"type": "enabled", "budget_tokens": 2000}}

    retry_count = 0
    while retry_count < max_retries:
        try:
            completion = client.chat.completions.create(
                model=model_name,
                messages=[{"role": "user", "content": prompt}],
                max_tokens=max_tokens,
                extra_headers=extra_headers,
                extra_body=extra_body,
            )
            return completion
        except Exception as e:
            retry_count += 1
            if retry_count >= max_retries:
                raise Exception(f"Failed after {max_retries} attempts. Last error: {str(e)}")

            # Exponential backoff with jitter
            delay = (2**retry_count) * 0.1 + (random.random() * 0.1)
            print(
                f"Request failed with error: {str(e)}. Retrying in {delay:.2f} seconds... (Attempt {retry_count}/{max_retries})"
            )
            time.sleep(delay)