BerriAI/litellm · error · BedrockError

Error setting response content: {e}. Response: {completion_r

Error message

Error setting response content: {e}. Response: {completion_response}

What it means

Raised by the TwelveLabs Pegasus transform_response when it cannot assign the extracted message content onto model_response.choices[0].message - either the content was empty/None, or tool_calls already exist on the message (making content assignment invalid). It wraps this as a BedrockError echoing the full completion_response and the underlying exception, using the HTTP status code of the raw response.

Source

Thrown at litellm/llms/bedrock/chat/invoke_transformations/amazon_twelvelabs_pegasus_transformation.py:239

        message_content: Final = completion_response.get("message", "")

        # Extract finish reason and map to LiteLLM format
        finish_reason_raw: Final = completion_response.get("finishReason", "stop")
        finish_reason: Final = map_finish_reason(finish_reason_raw)

        # Set the response content
        try:
            if (
                message_content
                and hasattr(model_response.choices[0], "message")
                and getattr(model_response.choices[0].message, "tool_calls", None) is None
            ):
                model_response.choices[0].message.content = message_content
                model_response.choices[0].finish_reason = finish_reason
            else:
                raise Exception("Unable to set message content")
        except Exception as e:
            raise BedrockError(
                message=f"Error setting response content: {e}. Response: {completion_response}",
                status_code=raw_response.status_code,
            )

        # Calculate usage from headers
        bedrock_input_tokens: Final = raw_response.headers.get("x-amzn-bedrock-input-token-count", None)
        bedrock_output_tokens: Final = raw_response.headers.get("x-amzn-bedrock-output-token-count", None)

        prompt_tokens: Final = int(bedrock_input_tokens or litellm.token_counter(messages=messages))

        completion_tokens: Final = int(
            bedrock_output_tokens
            or litellm.token_counter(
                text=model_response.choices[0].message.content,
                count_response_tokens=True,
            )
        )

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Inspect completion_response in the error message: an empty 'message' field means the model produced no content - check prompt and parameters.
2. Avoid passing tool definitions to twelvelabs-pegasus through this legacy invoke path; use the converse route if tool use is required.
3. Upgrade litellm so the newest twelvelabs transformation handles edge cases.
4. Verify max_tokens and stop settings are not truncating the response before any content is generated.

Example fix

# before
resp = litellm.completion(model="bedrock/us.twelvelabs-pegasus-1-0", messages=msgs, tools=tools)

# after - do not send tools to twelvelabs via the invoke path
resp = litellm.completion(model="bedrock/us.twelvelabs-pegasus-1-0", messages=msgs)

messages = [m for m in messages if m.get("content")]  # avoid empty-prompt edge cases
if max_tokens is not None and max_tokens < 8:
    raise ValueError("max_tokens too small - model may return empty message content")

from litellm.exceptions import BedrockError
try:
    resp = litellm.completion(model="bedrock/us.twelvelabs-pegasus-1-0", messages=msgs)
except BedrockError as e:
    if "Error setting response content" in str(e):
        log.warning("empty twelvelabs response: %s", e.message)
        return fallback_response()
    raise

When it happens

Trigger: A twelvelabs-pegasus completion where the JSON parses but the 'message' field is empty, or a previous tool-call path left choices[0].message.tool_calls set so the content branch is skipped and the explicit 'Unable to set message content' exception is thrown.

Common situations: Calling twelvelabs models with tools enabled (tool_calls present), provider returning an empty message field (e.g. content filtered or max_tokens=0), or version drift in the twelvelabs response schema.

Related errors

• Error parsing response: {raw_response.text}, error: {e}
• Error parsing received text={outputText}.\nError-{e}
• Model needs to be set for bedrock
• BedrockException: Context Window Error - {error_str}
• BedrockException - {error_str} . Enable 'litellm.modify_para

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/a77072427cecf528. Report an issue: GitHub.