langchain-ai/langchain · error · ValueError

Image URL not a data URI; appending as non-standard block.

Error message

Image URL not a data URI; appending as non-standard block.

What it means

Raised by the Google GenAI content block translator when an image item carries a URL that is not a `data:` URI. The translator can only inline images for Gemini as base64 data URIs; a remote http(s) URL cannot be passed through this path, so the block is recorded as non-standard and a ValueError is raised.

Source

Thrown at libs/core/langchain_core/messages/block_translators/google_genai.py:425

                                if mime_type:
                                    image_url_b64_block["mime_type"] = mime_type

                            converted_blocks.append(
                                cast("types.ImageContentBlock", image_url_b64_block)
                            )
                        except Exception:
                            # Not valid base64, treat as non-standard
                            converted_blocks.append(
                                {
                                    "type": "non_standard",
                                    "value": item,
                                }
                            )
                else:
                    # This likely won't be reached according to previous implementations
                    converted_blocks.append({"type": "non_standard", "value": item})
                    msg = "Image URL not a data URI; appending as non-standard block."
                    raise ValueError(msg)
            elif item_type == "function_call":
                # Handle Google GenAI function calls
                function_call_block: types.ToolCall = {
                    "type": "tool_call",
                    "name": item.get("name", ""),
                    "args": item.get("args", {}),
                    "id": item.get("id", ""),
                }
                converted_blocks.append(function_call_block)
            elif item_type == "file_data":
                # Handle FileData URI-based content
                file_block: types.FileContentBlock = {
                    "type": "file",
                    "url": item.get("file_uri", ""),
                }
                if mime_type := item.get("mime_type"):
                    file_block["mime_type"] = mime_type
                converted_blocks.append(file_block)

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Download the image and pass it as a data URI: `f"data:{mime};base64,{b64data}"`
2. Or upload the file to the Gemini Files API and pass a `file_data` block with the resulting file URI
3. If you control the block dict, set `mime_type` alongside the base64 data so the base64 branch is taken instead

Example fix

# before
block = {"type": "image_url", "image_url": {"url": "https://example.com/cat.png"}}

# after
import base64, mimetypes
raw = open("cat.png", "rb").read()
b64 = base64.b64encode(raw).decode()
block = {"type": "image_url", "image_url": {"url": f"data:{mimetypes.guess_type('cat.png')[0]};base64,{b64}"}}

def is_gemini_safe_image(block: dict) -> bool:
    url = block.get("url") or block.get("image_url", {}).get("url", "")
    return url.startswith("data:")

def is_data_uri_image(block: dict) -> bool:
    url = block.get("url") or block.get("image_url", {}).get("url", "")
    return isinstance(url, str) and url.startswith("data:")

try:
    resp = gemini_model.invoke([msg])
except ValueError as e:
    if "Image URL not a data URI" in str(e):
        msg = inline_image_as_data_uri(msg)  # download + base64 encode, then retry
    else:
        raise

When it happens

Trigger: Sending an AIMessage/content block to the google-genai path whose image entry is `{"type": "image_url", "image_url": {"url": "https://..."}}` or a plain dict with an `https://` URL, where the URL is not a `data:image/...;base64,...` URI and no file_data/mime_type alternative matched.

Common situations: Copying OpenAI-style image URL blocks into Gemini calls; hotlinking a CDN image instead of uploading via the Gemini Files API; assuming Gemini fetches remote image URLs like the OpenAI image_url parameter does.

Related errors

• mime_type key is required for base64 data.
• Unsupported source type. Only 'url' and 'base64' are support
• Must provide one of: url, base64, or file_id
• Keys base64, url, or file_id required for file blocks.
• Key base64 is required for audio blocks.

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/71fecd5668fe4c81. Report an issue: GitHub.