Hi there - I have the following python app that uses MockGPT:

import os
from openai import OpenAI
from flask import Flask, request
from opentelemetry.instrumentation.openai import OpenAIInstrumentor

app = Flask(__name__)

OpenAIInstrumentor().instrument()

client = OpenAI(
   api_key=os.environ.get("MOCK_GPT_API_KEY"),
   base_url="<https://mockgpt.wiremockapi.cloud/v1>"
)

@app.route("/askquestion", methods=['POST'])
def ask_question():
   data = request.json
   question = data.get('question')

   completion = client.chat.completions.create(
      model="gpt-3.5-turbo",
      messages=[
          {"role": "user", "content": question}
      ]
   )

   return completion.choices[0].message.content

It used to work fine, but now I encounter the following error:

Failed to encode key gen_ai.response.model: Invalid type <class 'NoneType'> of value None
Traceback (most recent call last):
  File "/Users/derekmitchell/Temp/openai-test/openai-env/lib/python3.9/site-packages/opentelemetry/exporter/otlp/proto/common/_internal/__init__.py", line 111, in _encode_attributes
    pb2_attributes.append(_encode_key_value(key, value))
  File "/Users/derekmitchell/Temp/openai-test/openai-env/lib/python3.9/site-packages/opentelemetry/exporter/otlp/proto/common/_internal/__init__.py", line 92, in _encode_key_value
    return PB2KeyValue(key=key, value=_encode_value(value))
  File "/Users/derekmitchell/Temp/openai-test/openai-env/lib/python3.9/site-packages/opentelemetry/exporter/otlp/proto/common/_internal/__init__.py", line 88, in _encode_value
    raise Exception(f"Invalid type {type(value)} of value {value}")
Exception: Invalid type <class 'NoneType'> of value None

The problem is that MockGPT is returning a model of "None" in the response, even though I set the model to "gpt-3.5-turbo" in the request, which is causing issues with OpenTelemetry instrumentation:

ChatCompletion(id='chatcmpl-123', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='Hello!\n\nThis is the default MockGPT response.\n\nCreate your own version in WireMock Cloud to fully customise this mock API.\n\n', refusal=None, role='assistant', function_call=None, tool_calls=None))], created=1728579960, model='gpt-3.5-turbo', object='chat.completion', service_tier=None, system_fingerprint=None, usage=CompletionUsage(completion_tokens=12, prompt_tokens=9, total_tokens=21))
127.0.0.1 - - [10/Oct/2024 10:06:00] "POST /askquestion HTTP/1.1" 200 -

In comparison, the same request sent to OpenAI's API returns a model:

ChatCompletion(id='chatcmpl-AGqfIFIyJzEqISoGz46VEsyfYLrLe', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='Hello! How can I assist you today?', role='assistant', function_call=None, tool_calls=None, refusal=None))], created=1728579056, model='gpt-3.5-turbo-0125', object='chat.completion', system_fingerprint=None, usage=CompletionUsage(completion_tokens=9, prompt_tokens=10, total_tokens=19, prompt_tokens_details={'cached_tokens': 0}, completion_tokens_details={'reasoning_tokens': 0}))

Could you please confirm if this behavior has changed, and if so, is it possible to return the model in the response again as the OpenAI API does?

Hi Derek, we haven’t updated this mock particularly recently so I’m wondering if this is a backwards-incompatible change in the Python libraries. What version of the OpenAI SDK and instrumentor are you running?

Hi Tom, thanks for looking into this. I've reverted back to the versions I used earlier, but that didn't seem to help. The versions I'm using are:

openai 1.43.0
opentelemetry-instrumentation-openai 0.28.2

Thanks for sharing that. Will try to take a look at this during the week.