self.status_code = int(original_status) if original_status is not None else 503

---

from litellm import CustomLLM, RateLimitError

class MyHandler(CustomLLM):
    async def acompletion(self, *args, **kwargs):
        raise RateLimitError(
            message="Key exhausted",
            model="my-model",
            llm_provider="openai",
        )

    async def astreaming(self, *args, **kwargs):
        # This triggers the bug: RateLimitError propagates through
        # the streaming pipeline and hits MidStreamFallbackError
        await self.acompletion(*args, **kwargs)
        yield  # never reached

---

model_list:
  - model_name: my-model
    litellm_params:
      model: my-provider/my-model
  - model_name: fallback-model
    litellm_params:
      model: openai/gpt-4

litellm_settings:
  custom_provider_map:
    - {"provider": "my-provider", "custom_handler": my_handler}
  fallbacks:
    - my-model:
        - fallback-model

---

File "litellm/exceptions.py", line 957, in __init__
    self.status_code = int(original_status) if original_status is not None else 503
ValueError: invalid literal for int() with base 10: 'litellm_error'

---

File "litellm/litellm_core_utils/streaming_handler.py", line 2181, in __anext__
    self._handle_stream_fallback_error(e)
File "litellm/litellm_core_utils/streaming_handler.py", line 2249, in _handle_stream_fallback_error
    raise MidStreamFallbackError(
        message=str(mapped_exception),
        ...
        is_pre_first_chunk=not self.sent_first_chunk,
    )
File "litellm/exceptions.py", line 957, in __init__
    self.status_code = int(original_status) if original_status is not None else 503
ValueError: invalid literal for int() with base 10: 'litellm_error'

---

# Before (crashes):
self.status_code = int(original_status) if original_status is not None else 503

# After (safe):
try:
    self.status_code = int(original_status) if original_status is not None else 503
except (ValueError, TypeError):
    self.status_code = 503

---