[Bug]: Assertion Error while using google models via OpenRouter

[kushalsharma]

What happened?

Code to reproduce bug :

from litellm import completion
import os

os.environ['OPENROUTER_API_KEY'] = ""
response = completion(
    model="openrouter/google/gemini-flash-1.5", 
    messages=[{"role": "user", "content": "some NSFW content to trigger error"}]
)

Relevant log output

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(response_object, model_response_object, response_type, stream, start_time, end_time, hidden_params)
   5321             return model_response_object
   5322     except Exception as e:
-> 5323         raise Exception(
   5324             f"Invalid response object {traceback.format_exc()}\n\nreceived_args={received_args}"

AssertionError: 

During handling of the above exception, another exception occurred:

Exception                                 Traceback (most recent call last)
~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/llms/openai.py in ?(self, model_response, timeout, optional_params, model, messages, print_verbose, api_key, api_base, acompletion, logging_obj, litellm_params, logger_fn, headers, custom_prompt_dict, client, organization, custom_llm_provider)
    831                 raise OpenAIError(status_code=e.status_code, message=str(e))
    832             else:
--> 833                 raise OpenAIError(status_code=500, message=traceback.format_exc())

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/llms/openai.py in ?(self, model_response, timeout, optional_params, model, messages, print_verbose, api_key, api_base, acompletion, logging_obj, litellm_params, logger_fn, headers, custom_prompt_dict, client, organization, custom_llm_provider)
    831                 raise OpenAIError(status_code=e.status_code, message=str(e))
    832             else:
--> 833                 raise OpenAIError(status_code=500, message=traceback.format_exc())

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(response_object, model_response_object, response_type, stream, start_time, end_time, hidden_params)
   5321             return model_response_object
   5322     except Exception as e:
-> 5323         raise Exception(
   5324             f"Invalid response object {traceback.format_exc()}\n\nreceived_args={received_args}"

Exception: Invalid response object Traceback (most recent call last):
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py", line 5198, in convert_to_model_response_object
    assert response_object["choices"] is not None and isinstance(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError


received_args={'response_object': {'id': None, 'choices': None, 'created': None, 'model': None, 'object': None, 'service_tier': None, 'system_fingerprint': None, 'usage': None, 'error': {'message': "Cannot read properties of undefined (reading 'parts')", 'code': 502}}, 'model_response_object': ModelResponse(id='chatcmpl-86f2f5d4-daf7-4be1-8520-fd5d02e56577', choices=[Choices(finish_reason='stop', index=0, message=Message(content='default', role='assistant'))], created=1719056210, model='None/google/gemini-flash-1.5', object='chat.completion', system_fingerprint=None, usage=Usage()), 'response_type': 'completion', 'stream': False, 'start_time': None, 'end_time': None, 'hidden_params': None}

During handling of the above exception, another exception occurred:

OpenAIError                               Traceback (most recent call last)
File ~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/main.py:1791, in completion(model, messages, timeout, temperature, top_p, n, stream, stream_options, stop, max_tokens, presence_penalty, frequency_penalty, logit_bias, user, response_format, seed, tools, tool_choice, logprobs, top_logprobs, deployment_id, extra_headers, functions, function_call, base_url, api_version, api_key, model_list, **kwargs)
   1790 ## COMPLETION CALL
-> 1791 response = openai_chat_completions.completion(
   1792     model=model,
   1793     messages=messages,
   1794     headers=headers,
   1795     api_key=api_key,
   1796     api_base=api_base,
   1797     model_response=model_response,
   1798     print_verbose=print_verbose,
   1799     optional_params=optional_params,
   1800     litellm_params=litellm_params,
   1801     logger_fn=logger_fn,
   1802     logging_obj=logging,
   1803     acompletion=acompletion,
   1804     timeout=timeout,  # type: ignore
   1805 )
   1806 ## LOGGING

File ~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/llms/openai.py:833, in OpenAIChatCompletion.completion(self, model_response, timeout, optional_params, model, messages, print_verbose, api_key, api_base, acompletion, logging_obj, litellm_params, logger_fn, headers, custom_prompt_dict, client, organization, custom_llm_provider)
    832 else:
--> 833     raise OpenAIError(status_code=500, message=traceback.format_exc())

OpenAIError: Traceback (most recent call last):
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py", line 5198, in convert_to_model_response_object
    assert response_object["choices"] is not None and isinstance(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/llms/openai.py", line 825, in completion
    raise e
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/llms/openai.py", line 792, in completion
    return convert_to_model_response_object(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py", line 5323, in convert_to_model_response_object
    raise Exception(
Exception: Invalid response object Traceback (most recent call last):
  File "/Users/kush/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py", line 5198, in convert_to_model_response_object
    assert response_object["choices"] is not None and isinstance(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError


received_args={'response_object': {'id': None, 'choices': None, 'created': None, 'model': None, 'object': None, 'service_tier': None, 'system_fingerprint': None, 'usage': None, 'error': {'message': "Cannot read properties of undefined (reading 'parts')", 'code': 502}}, 'model_response_object': ModelResponse(id='chatcmpl-86f2f5d4-daf7-4be1-8520-fd5d02e56577', choices=[Choices(finish_reason='stop', index=0, message=Message(content='default', role='assistant'))], created=1719056210, model='None/google/gemini-flash-1.5', object='chat.completion', system_fingerprint=None, usage=Usage()), 'response_type': 'completion', 'stream': False, 'start_time': None, 'end_time': None, 'hidden_params': None}


During handling of the above exception, another exception occurred:

UnboundLocalError                         Traceback (most recent call last)
/var/folders/_0/5yzypzd977xb2k9p8ljjxmm00000gn/T/ipykernel_68268/1428737860.py in ?()
      1 from litellm import completion
      2 import os
      3 
      4 os.environ['OPENROUTER_API_KEY'] = "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"
----> 5 response = completion(
      6     model="openrouter/google/gemini-flash-1.5",
      7     messages=[{"role": "user", "content": "some text to generate NSFW content"}]
      8 )

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(*args, **kwargs)
    951                     if (
    952                         liteDebuggerClient and liteDebuggerClient.dashboard_url != None
    953                     ):  # make it easy to get to the debugger logs if you've initialized it
    954                         e.message += f"\n Check the log in your dashboard - {liteDebuggerClient.dashboard_url}"
--> 955             raise e

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(*args, **kwargs)
    951                     if (
    952                         liteDebuggerClient and liteDebuggerClient.dashboard_url != None
    953                     ):  # make it easy to get to the debugger logs if you've initialized it
    954                         e.message += f"\n Check the log in your dashboard - {liteDebuggerClient.dashboard_url}"
--> 955             raise e

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/main.py in ?(model, messages, timeout, temperature, top_p, n, stream, stream_options, stop, max_tokens, presence_penalty, frequency_penalty, logit_bias, user, response_format, seed, tools, tool_choice, logprobs, top_logprobs, deployment_id, extra_headers, functions, function_call, base_url, api_version, api_key, model_list, **kwargs)
   2576             )
   2577         return response
   2578     except Exception as e:
   2579         ## Map to OpenAI Exception
-> 2580         raise exception_type(
   2581             model=model,
   2582             custom_llm_provider=custom_llm_provider,
   2583             original_exception=e,

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(model, original_exception, custom_llm_provider, completion_kwargs, extra_kwargs)
   7436         ):
   7437             threading.Thread(target=get_all_keys, args=(e.llm_provider,)).start()
   7438         # don't let an error with mapping interrupt the user from receiving an error from the llm api calls
   7439         if exception_mapping_worked:
-> 7440             raise e
   7441         else:
   7442             raise APIConnectionError(
   7443                 message="{}\n{}".format(original_exception, traceback.format_exc()),

~/miniforge3/envs/mm-scripts/lib/python3.12/site-packages/litellm/utils.py in ?(model, original_exception, custom_llm_provider, completion_kwargs, extra_kwargs)
   7436         ):
   7437             threading.Thread(target=get_all_keys, args=(e.llm_provider,)).start()
   7438         # don't let an error with mapping interrupt the user from receiving an error from the llm api calls
   7439         if exception_mapping_worked:
-> 7440             raise e
   7441         else:
   7442             raise APIConnectionError(
   7443                 message="{}\n{}".format(original_exception, traceback.format_exc()),

UnboundLocalError: cannot access local variable 'exception_provider' where it is not associated with a value

Twitter / LinkedIn details

No response