BerriAI/litellm · error · ValueError

Unsupported image format: {image_format}. Supported formats:

Error message

Unsupported image format: {image_format}. Supported formats: {supported_image_and_video_formats}

What it means

Bedrock image validation: for non-document payloads, the resolved image_format (extension like 'jpg', 'gif', 'mp4') is checked against the union of supported image and video formats, and this ValueError is raised when it is not in that list. The supported list is embedded in the message.

Source

Thrown at litellm/litellm_core_utils/prompt_templates/factory.py:3473

        supported_doc_formats: Final = litellm.AmazonConverseConfig().get_supported_document_types()
        supported_video_formats: Final = litellm.AmazonConverseConfig().get_supported_video_types()

        document_types: Final = ["application", "text"]
        is_document: Final = any(mime_type.startswith(doc_type) for doc_type in document_types)

        supported_image_and_video_formats: Final[list[str]] = supported_video_formats + supported_image_formats

        if is_document:
            return BedrockImageProcessor._get_document_format(
                mime_type=mime_type, supported_doc_formats=supported_doc_formats
            )

        else:
            #########################################################
            # Check if image_format is an image or video
            #########################################################
            if image_format not in supported_image_and_video_formats:
                raise ValueError(
                    f"Unsupported image format: {image_format}. Supported formats: {supported_image_and_video_formats}"
                )
            return image_format

    @staticmethod
    def _get_document_format(mime_type: str, supported_doc_formats: list[str]) -> str:
        """
        Get the document format from the mime type

        - Primary method - uses `mimetypes.guess_all_extensions`
        - Fallback method - uses `get_file_extension_from_mime_type`

        Relevant Issue: https://github.com/BerriAI/litellm/issues/12260

        `mimetypes` is not available in docker containers, so we fallback to `get_file_extension_from_mime_type`

        Args:
            mime_type: The mime type of the document

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Convert to a supported format first (jpeg/png/gif/webp for images; Bedrock-listed video formats) with Pillow/ffmpeg
2. Check the error's supported list and re-encode your asset to one of those extensions/mime types
3. Ensure the mime_type you pass matches the actual bytes so the correct format is derived

Example fix

# before
block = {"type": "image_url", "image_url": {"url": "data:image/tiff;base64,<b64>"}}

# after
from PIL import Image
import io, base64
img = Image.open("scan.tiff").convert("RGB")
buf = io.BytesIO(); img.save(buf, format="JPEG")
block = {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64," + base64.b64encode(buf.getvalue()).decode()}}

SUPPORTED_IMAGE_FORMATS = {"jpeg", "jpg", "png", "gif", "webp"}
# cross-check with the list in the error message for video formats on your Bedrock model

def validate_bedrock_image(blocks) -> None:
    for b in blocks:
        if isinstance(b, dict) and b.get("type") == "image_url":
            fmt = image_format_from_url(b["image_url"]["url"])  # e.g. from data:image/<fmt>
            if fmt.lower() not in SUPPORTED_IMAGE_FORMATS:
                raise ValueError(f"convert {fmt} to a supported format before Bedrock")

from typing import Any

_OK = {"jpeg", "jpg", "png", "gif", "webp"}

def is_bedrock_supported_image(v: Any) -> bool:
    if not isinstance(v, str):
        return False
    if v.startswith("data:image/"):
        fmt = v.split("data:image/")[1].split(";", 1)[0]
        return fmt.lower() in _OK
    return False

When it happens

Trigger: Sending a Bedrock converse/vision request with an image whose format is not one of the supported image/video extensions (e.g. tiff, heic, bmp, svg); a format inferred from a mime type that maps to an unsupported extension; video formats outside Bedrock's accepted set.

Common situations: Feeding raw camera/scanner output (TIFF/HEIC) to Bedrock models; mislabelled mime types causing wrong extension inference; assuming S3-stored proprietary formats will pass through.

Related errors

• Invalid image URL: {content_image_url}
• Image url not in expected format. Example Expected input - "
• No supported extensions for MIME type: {mime_type}. Supporte
• Invalid guardrailConfig={raw_guardrail_config}. Expected for
• guardrailConfig={raw_guardrail_config} is missing 'guardrail

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/2291ab1b5c72cdd9. Report an issue: GitHub.