langchain-ai/langchain · error · ValueError

mime_type key is required for base64 data.

Error message

mime_type key is required for base64 data.

What it means

Raised when converting a standard image content block to the OpenAI format and the block indicates base64 data (`base64` key or `source_type == "base64"`) but has no `mime_type` key. OpenAI requires a complete data URI (`data:<mime>;base64,<data>`), which cannot be built without a MIME type.

Source

Thrown at libs/core/langchain_core/messages/block_translators/openai.py:45

    Raises:
        ValueError: If required keys are missing.
        ValueError: If source type is unsupported.

    Returns:
        The formatted image content block.
    """
    if "url" in block:
        return {
            "type": "image_url",
            "image_url": {
                "url": block["url"],
            },
        }
    if "base64" in block or block.get("source_type") == "base64":
        if "mime_type" not in block:
            error_message = "mime_type key is required for base64 data."
            raise ValueError(error_message)
        mime_type = block["mime_type"]
        base64_data = block["data"] if "data" in block else block["base64"]
        return {
            "type": "image_url",
            "image_url": {
                "url": f"data:{mime_type};base64,{base64_data}",
            },
        }
    error_message = "Unsupported source type. Only 'url' and 'base64' are supported."
    raise ValueError(error_message)


def convert_to_openai_data_block(
    block: dict[str, Any],
    api: Literal["chat/completions", "responses"] = "chat/completions",
) -> dict[str, Any]:
    """Format standard data content block to format expected by OpenAI.

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Add `mime_type` to the block, e.g. `{..., "mime_type": "image/png"}`
2. Build blocks through the v1 helper which enforces the pair: `ImageContentBlock(base64=..., mime_type="image/png")`
3. Infer the MIME type from a file extension or magic bytes before sending: `mimetypes.guess_type(path)[0]`

Example fix

# before
block = {"type": "image", "base64": b64_data}

# after
block = {"type": "image", "base64": b64_data, "mime_type": "image/png"}

def valid_openai_image_block(block: dict) -> bool:
    if "url" in block:
        return True
    if "base64" in block or block.get("source_type") == "base64":
        return "mime_type" in block
    return False

def is_b64_image_with_mime(block: dict) -> bool:
    return ("base64" in block or block.get("source_type") == "base64") and "mime_type" in block

When it happens

Trigger: Passing a block like `{"type": "image", "source_type": "base64", "data": "..."}` (v0 style) or `{"type": "image", "base64": "..."}` (v1 style) without a `mime_type` field to an OpenAI model or to `convert_to_openai_data_block`.

Common situations: Hand-building image blocks from raw bytes and forgetting the MIME type; blocks produced by another provider's output that never carried `mime_type`; migrating from integrations that inferred the MIME type from context.

Related errors

• Unsupported source type. Only 'url' and 'base64' are support
• Key base64 is required for audio blocks.
• mime_type is required when using base64 data
• Image URL not a data URI; appending as non-standard block.
• Keys base64, url, or file_id required for file blocks.

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/82a87bcc1b005258. Report an issue: GitHub.