BerriAI/litellm · error · HTTPException

Failed to parse Braintrust API response: {str(e)}

Error message

Failed to parse Braintrust API response: {str(e)}

What it means

A guard inside OpenAI text-completion-to-chat response conversion: when converting a raw TextCompletionResponse into LiteLLM's ModelResponse format, the transform requires both a non-None response_object and model_response_object. If either is None (e.g. the caller passed no response dict, or an internal path failed to initialize the target ModelResponse), a ValueError with the generic message 'Error in response object format' is raised. It indicates a malformed or missing payload rather than an HTTP failure.

Source

Thrown at cookbook/litellm_proxy_server/braintrust_prompt_wrapper_server.py:221

    print(f"braintrust_token: {braintrust_token}")

    try:
        async with httpx.AsyncClient(timeout=30.0) as client:
            response = await client.get(braintrust_url, headers=headers)
            response.raise_for_status()
            braintrust_data = response.json()
    except httpx.HTTPStatusError as e:
        raise HTTPException(
            status_code=e.response.status_code,
            detail=f"Braintrust API error: {e.response.text}",
        )
    except httpx.RequestError as e:
        raise HTTPException(
            status_code=502,
            detail=f"Failed to connect to Braintrust API: {str(e)}",
        )
    except json.JSONDecodeError as e:
        raise HTTPException(
            status_code=502,
            detail=f"Failed to parse Braintrust API response: {str(e)}",
        )

    print(f"braintrust_data: {braintrust_data}")
    # Transform the response
    try:
        transformed_data = transform_braintrust_response(braintrust_data)
        print(f"transformed_data: {transformed_data}")
        return JSONResponse(content=transformed_data)
    except Exception as e:
        raise HTTPException(
            status_code=500,
            detail=f"Failed to transform Braintrust response: {str(e)}",
        )


@app.get("/health")

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Log the raw HTTP response body (set LITELLM_LOG=DEBUG) to see what the provider actually returned.
2. If using a custom api_base/proxy, verify it returns an OpenAI-spec completions JSON body with a choices array.
3. Reproduce with litellm.text_completion() to confirm the raw path works, isolating the conversion layer.
4. Upgrade litellm to the latest patch release in case a provider response-schema change was fixed.

Example fix

// not a caller-fixable error; capture the raw payload for diagnosis
# before
resp = litellm.completion(model="openai/gpt-3.5-turbo-instruct", messages=[...])

# after
import litellm, logging
litellm.suppress_debug_info = True
logging.basicConfig(level=logging.DEBUG)
resp = litellm.completion(model="openai/gpt-3.5-turbo-instruct", messages=[...])  # inspect raw body in debug logs

def is_valid_text_completion_body(body: dict) -> bool:
    return isinstance(body, dict) and isinstance(body.get("choices"), list) and len(body["choices"]) > 0

try:
    resp = litellm.completion(model="openai/gpt-3.5-turbo-instruct", messages=[...])
except ValueError as e:
    if "Error in response object format" in str(e):
        logger.error("provider returned malformed/empty body; capture raw response with LITELLM_LOG=DEBUG")
    raise

When it happens

Trigger: Calling litellm.completion() on a text-completion model where the transformation layer receives a None response object (empty provider response body) or an uninitialized ModelResponse; typically after a 200 response whose body failed to deserialize into TextCompletionResponse.

Common situations: Custom api_base gateways returning empty 200 bodies; provider API changes altering response shape; mocking/streaming tests that pass None where a response object is expected; version mismatches between litellm and the openai SDK pydantic models.

Understand the failure class

• Parsing and encoding errors: unexpected token, malformed input — why parsers reject input and how to find the real culprit.

Related errors

• Braintrust API error: {e.response.text}
• Failed to connect to Braintrust API: {str(e)}
• Error apply_db_fixes: {str(e)}
• Unclassified keys in {PRICES_PATH.name}: {', '.join(unclassi
• No Braintrust API token provided. Pass via Authorization hea

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/a42feb5a69475a6e. Report an issue: GitHub.