litellm - 💡(How to fix) Fix fix(github_copilot): gpt-5.4 requires Responses API but proxy routes through /chat/completions [1 comments, 1 participants]

Fix Plan

The fix involves modifying the LiteLLM proxy to use the correct endpoint for the gpt-5.4 model.

• Update the LiteLLM proxy to use the /v1/completions endpoint for gpt-5.4 model requests.
• Add a conditional statement to check the model type and route the request accordingly.

Example code:

if model == "gpt-5.4":
    # Use the /v1/completions endpoint for gpt-5.4 model
    response = litellm.completion(prompt, model="gpt-5.4", endpoint="/v1/completions")
else:
    # Use the /chat/completions endpoint for other models
    response = litellm.completion(prompt, model=model, endpoint="/chat/completions")

Verification

To verify the fix, send a request through the LiteLLM proxy with the gpt-5.4 model and check if the response is successful.

Extra Tips

• Make sure to update the LiteLLM proxy documentation to reflect the changes.
• Test the fix with different models and endpoints to ensure compatibility.