ATH-MaaS/Pixelle-Video · error · RuntimeError

OpenAI API 返回数据为空

Error message

OpenAI API 返回数据为空

What it means

generate_image calls the OpenAI Images API (via client.images.generate with extra_body support) and raises RuntimeError when the SDK returns None or an empty data list. This means the relay/proxy or OpenAI returned a success-shaped response with no image payloads.

Source

Thrown at pixelle_video/services/api_services/image_gpt.py:101

        if image_urls and isinstance(image_urls, list) and len(image_urls) > 0:
            # 中转站通常支持通过 extra_body 传递 image_url 或 ref_image
            # 这里我们将第一张图作为参考图
            ref_images = [self._encode_image_to_base64(image_urls[i]) for i in range(min(len(image_urls), 6))]
            extra_body = {"image_url": ref_images}

        while attempts < self.max_attempts:
            try:
                response = self.client.images.generate(
                    model=model,
                    prompt=prompt,
                    size=size,
                    quality=quality,
                    n=1,
                    extra_body=extra_body
                )
                
                if not response or not response.data:
                    raise RuntimeError("OpenAI API 返回数据为空")

                img_data = response.data[0]
                file_path = None

                # 1. 处理 Base64 格式 (中转站常用)
                if hasattr(img_data, 'b64_json') and img_data.b64_json:
                    if save_dir:
                        os.makedirs(save_dir, exist_ok=True)
                        file_name = f"gpt_{int(time.time())}_{uuid.uuid4().hex[:6]}.png"
                        file_path = os.path.join(save_dir, file_name)
                        with open(file_path, "wb") as f:
                            f.write(base64.b64decode(img_data.b64_json))
                        return file_path
                    return img_data.b64_json

                # 2. 处理 URL 格式
                elif hasattr(img_data, 'url') and img_data.url:
                    url = img_data.url

View on GitHub (pinned to 848b054e4f)

Solutions

1. Retry the request — this library already retries per model; transient relay failures usually resolve on retry.
2. Verify base_url points to a working endpoint and the API key is valid for that endpoint.
3. Try a different supported model (e.g. 'gpt-image-1' vs 'dall-e-3') via the model argument.
4. Check the relay provider's status/logs; switch to the official API to rule out relay issues.

Example fix

// before
url = client.generate_image(prompt, model="dall-e-3")

// after
from tenacity import retry, stop_after_attempt
@retry(stop=stop_after_attempt(3), wait=wait_fixed(10))
def gen():
    return client.generate_image(prompt, model="gpt-image-1")

assert api_key and api_key.startswith(("sk-",)), "OpenAI/relay key missing"
assert base_url, "base_url for OpenAI-compatible endpoint not configured"

def has_image_data(resp) -> bool:
    return resp is not None and bool(getattr(resp, "data", None))

try:
    out = client.generate_image(prompt)
except RuntimeError as e:
    if "返回数据为空" in str(e):
        time.sleep(10)
        out = client.generate_image(prompt)  # transient relay failure
    else:
        raise

When it happens

Trigger: client.images.generate(...) returns None, or response.data is an empty list — typically from a third-party relay/中转站 returning 200 with an empty body, or the model returning no images.

Common situations: Using an unofficial OpenAI-compatible relay that silently fails, base_url pointing to a proxy with intermittent empty responses, model name not supported by the endpoint, content policy rejection returning empty data instead of an error.

Related errors

• API image generation returned no result: provider={provider}
• DashScope image generation returned no image URLs. output={g
• 未在响应中找到 url 或 b64_json
• Max attempts reached, failed to generate image. Last error:
• str(e)

AI-assisted analysis of ATH-MaaS/Pixelle-Video@848b054e4f (2026-08-30). Data as JSON: /api/errors/9b127093f42ba013. Report an issue: GitHub.