Url problem in AsyncCompletions

[gmirc12]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• [x] This is an issue with the Python library

Describe the bug

In src/openai/resources/chat/completions/completions.py line 1928 in AsyncCompletions.create

(https://github.com/openai/openai-python/blob/d6bb8c14e66605ad2b7ed7bd62951014cd21b576/src/openai/resources/chat/completions/completions.py#L1928)

you set "/chat/completions" as a path variable, but you actually set it as a url, so the client tryes to hit a /chat/completions url without any baseUrl prepended to it.

To Reproduce

Error: openai.NotFoundError: Error code: 404 - {'error': {'code': '404', 'message': 'Resource not found'}}

Code snippets

from llama_index.llms.openai import OpenAI

llm = OpenAI(
            temperature=temp,
            max_tokens=tokens,
            model="model",
            api_key="key",
            api_base="baseUrl",
            default_headers = {headers}
        )
resp = llm.acomplete("Test")


#OR

from openai import OpenAI

client = OpenAI(
            api_key="key",
            base_url="url",
            default_headers = {headers}
        )
response = client.chat.completions.create(
          model="model",
          n=1,
          messages=[messages]
)

OS

WSL

Python version

v3.12

Library version

1.65.3

[abhinav-sreekumar]

Hi, I’d like to work on this issue. I’ll start investigating and update here if I have any questions.

[RobertCraigie]

Thanks for the report but I cannot reproduce this. Sounds like llama index is patching the client incorrectly?