BerriAI/litellm · error · Exception

File not found. banned_keywords_list={banned_keywords_list}

Error message

File not found. banned_keywords_list={banned_keywords_list}

What it means

In the OpenAI image-variations handler, the async preparation step (transforming the request via the provider transformation layer) is wrapped in a catch-all: any exception raised while building the request (param validation errors like 'Parameter ... is not supported', encoding failures of the input image, invalid API base) is re-raised as OpenAIError with the upstream status_code (default 500), headers, and text. The literal 'error_text' means the original exception had no .text attribute, so str(e) was used - the real cause is in the wrapped message.

Source

Thrown at enterprise/enterprise_hooks/banned_keywords.py:42

    # Class variables or attributes
    def __init__(self):
        banned_keywords_list = litellm.banned_keywords_list

        if banned_keywords_list is None:
            raise Exception(
                "`banned_keywords_list` can either be a list or filepath. None set."
            )

        if isinstance(banned_keywords_list, list):
            self.banned_keywords_list = banned_keywords_list

        if isinstance(banned_keywords_list, str):  # assume it's a filepath
            try:
                with open(banned_keywords_list, "r") as file:
                    data = file.read()
                    self.banned_keywords_list = data.split("\n")
            except FileNotFoundError:
                raise Exception(
                    f"File not found. banned_keywords_list={banned_keywords_list}"
                )
            except Exception as e:
                raise Exception(
                    f"An error occurred: {str(e)}, banned_keywords_list={banned_keywords_list}"
                )

    def print_verbose(self, print_statement, level: Literal["INFO", "DEBUG"] = "DEBUG"):
        if level == "INFO":
            verbose_proxy_logger.info(print_statement)
        elif level == "DEBUG":
            verbose_proxy_logger.debug(print_statement)

        if litellm.set_verbose is True:
            print(print_statement)  # noqa

    def test_violation(self, test_str: str):
        for word in self.banned_keywords_list:

View on GitHub (pinned to 6c2dcb801b)

Solutions

Catch OpenAIError and read .message - it contains the underlying transformation exception text.
Remove unsupported params or set drop_params=True.
Verify the input image is a valid, readable PNG and the model supports variations (dall-e-2).
Upgrade litellm if the transformation layer changed for your model.

Example fix

# before
img = await litellm.aimage_variations(model="dall-e-2", image=open("cat.png","rb"), n=2)

# after
try:
    img = await litellm.aimage_variations(model="dall-e-2", image=open("cat.png","rb"), n=2)
except litellm.exceptions.OpenAIError as e:
    logger.error("image_variations failed: %s", e.message)
    raise

Defensive patterns

Strategy: try-catch

Validate before calling

from pathlib import Path

def variation_input_ok(path: str) -> bool:
    p = Path(path)
    return p.exists() and p.stat().st_size > 0 and p.read_bytes()[:8] == b"\x89PNG\r\n\x1a\n"

Type guard

from litellm.exceptions import OpenAIError

def is_transform_layer_error(e: BaseException) -> bool:
    return isinstance(e, OpenAIError) and getattr(e, "status_code", 500) == 500 and e.headers is None

Try / catch

from litellm.exceptions import OpenAIError

try:
    img = await litellm.aimage_variations(model="dall-e-2", image=image_bytes, n=2)
except OpenAIError as e:
    logger.error("image_variations failed: %s", e.message)  # e.message holds the transform-layer cause
    raise

Prevention

Verify the input image is a valid PNG before calling (dall-e-2 variations require PNG).
Pass only params the model supports; prefer drop_params=True for user-driven param dicts.
Log e.message fully - the underlying ValueError text is nested inside the OpenAIError.

When it happens

Trigger: Calling litellm.image_variations() (async path) with an unsupported parameter for the model (which raises in transformation), a corrupted or wrong-format image file, an unwritable/missing api_base, or missing credentials that surface during request preparation rather than the HTTP call.

Common situations: Generating variations of dall-e-2 images with dall-e-3-only params; passing PNG bytes where the API expects a valid image the encoder can process; mismatched model names routed to the image_variations endpoint; transformation-layer version drift after litellm upgrades.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/612ff331cd08e9d3. Report an issue: GitHub.