BerriAI/litellm · error · BytezError

str(json["error"])

Error message

str(json["error"])

What it means

After a Bytez chat response arrives, the transformation looks for a top-level 'error' key in the JSON body; if present it raises BytezError with the HTTP status code and str() of the error value. This is how Bytez-side failures (bad model path, insufficient credits, model load problems) surface through litellm.

Source

Thrown at litellm/llms/bytez/chat/transformation.py:197

        self,
        model: str,
        raw_response: httpx.Response,
        model_response: ModelResponse,
        logging_obj: LiteLLMLoggingObj,
        request_data: dict,
        messages: list[AllMessageValues],
        optional_params: dict,
        litellm_params: dict,
        encoding: Any,
        api_key: str | None = None,
        json_mode: bool | None = None,
    ) -> ModelResponse:
        json: Final = raw_response.json()

        error: Final = json.get("error")

        if error is not None:
            raise BytezError(
                message=str(json["error"]),
                status_code=raw_response.status_code,
            )

        # set meta data here
        model_response.created = int(time.time())
        model_response.model = model

        # Add the output
        output: Final = json.get("output")

        message: Final = model_response.choices[0].message

        message.content = output["content"][0]["text"]

        messages = adapt_messages_to_bytez_standard(messages=messages)

        # NOTE We are approximating tokens, to get the true values we will need to update our BE

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Catch BytezError and read status_code + message — they carry the upstream reason verbatim.
2. Fix the model string (correct org/model path) or enable the model in your Bytez account.
3. Retry with backoff for load-related/transient errors (first-request cold start).
4. Check account credits/plan if the message indicates billing.

from litellm.llms.bytez.common_utils import BytezError

try:
    resp = litellm.completion(model="bytez/org/model", messages=msgs)
except BytezError as e:
    if e.status_code in (429, 500, 503):
        # transient: retry with backoff
        ...
    elif e.status_code in (401, 402, 403):
        raise RuntimeError(f"Bytez account/key problem: {e.message[:200]}") from e
    else:
        raise

When it happens

Trigger: Requesting a bytez model path that does not exist or is not enabled for your account; out-of-credit or unauthorized account; cold-start model still loading server-side (the 5-minute STREAMING_TIMEOUT comment notes models may need to load); malformed model identifiers.

Common situations: Wrong model org/name casing in the bytez model string; free-tier limits; first request to a rarely used model that must be loaded; shared keys rotated without updating config.

Related errors

• No audio part found in the response
• Mavvrik FOCUS destination: response missing 'url' field: {re
• ReplicateException - {original_exception}
• AzureException APIError - {message}
• APIError: {exception_provider} - {error_str}

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/e93e950f0767e6f5. Report an issue: GitHub.