microsoft/semantic-kernel · error · ServiceResponseException

No valid image data found in response.

Error message

No valid image data found in response.

What it means

Raised by generate_images after it iterates response.data collecting url/b64_json values. If data items existed but none carried a usable url or b64_json, results stays empty and it throws ServiceResponseException. This means the API returned image objects but every one lacked the expected payload fields.

Source

Thrown at python/semantic_kernel/connectors/ai/open_ai/services/open_ai_text_to_image_base.py:142

        response = await self._send_request(settings)

        assert isinstance(response, ImagesResponse)  # nosec
        if not response.data or not isinstance(response.data, list) or len(response.data) == 0:
            raise ServiceResponseException("Failed to generate image.")

        results: list[str] = []
        for image in response.data:
            url: str | None = getattr(image, "url", None)
            b64_json: str | None = getattr(image, "b64_json", None)
            if url:
                results.append(url)
            elif b64_json:
                results.append(b64_json)
            else:
                continue

        if len(results) == 0:
            raise ServiceResponseException("No valid image data found in response.")
        return results

    async def edit_image(
        self,
        prompt: str,
        image_paths: list[str] | None = None,
        image_files: list[IO[bytes]] | None = None,
        mask_path: str | None = None,
        mask_file: IO[bytes] | None = None,
        settings: PromptExecutionSettings | None = None,
        **kwargs: Any,
    ) -> list[str]:
        """Edit images using the OpenAI image edit API.

        Args:
            prompt: Instructional prompt for image editing.
            image_paths: List of image file paths to edit.
            image_files: List of file-like objects (opened in binary mode) to edit.

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Set response_format on settings to match the field you intend to consume ('b64_json' or 'url')
2. If using a compatible non-OpenAI endpoint, confirm it populates url/b64_json
3. Retry with a different prompt to rule out a content filter on the specific input

Example fix

# before
settings.response_format = "b64_json"
await service.generate_images(prompt="...", settings=settings)
# raises: No valid image data found in response.

# after
settings.response_format = "url"
await service.generate_images(prompt="...", settings=settings)

from semantic_kernel.exceptions.service_exceptions import ServiceResponseException

try:
    images = await service.generate_images(prompt=prompt, settings=settings)
except ServiceResponseException as e:
    if "No valid image data" in str(e):
        # data items lacked url/b64_json; adjust response_format and retry
        settings.response_format = "url"
        ...
    raise

When it happens

Trigger: response.data is a non-empty list, but each item's url and b64_json attributes are both None/empty. This can happen with an unexpected response_format, a content-filtered per-item result, or a non-conforming provider response.

Common situations: Requested b64_json but the endpoint returned only revised_prompt metadata; using a third-party-compatible endpoint that omits these fields; per-image filter with no fallback.

Related errors

• Failed to generate image.
• Failed to edit image.
• The OpenAI text to image model ID is required.
• Prompt is required.
• Provide either 'image_paths' or 'image_files', and only one.

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/0cdec437ebf708f3. Report an issue: GitHub.