18 views
-/https://github.com/berriai/litellm/issues/26309
GitHub · issue

#26309 [Bug]: `chatgpt/gpt-5.4` throw exception when stream is `false`

  • State: open
  • Author: @KAIYOHUGO
  • Labels: bug, proxy

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

The `chatgpt/gpt-5.4` throw exception when stream is `false`. It work fine with other `chatgpt/*` model.

### Steps to Reproduce

1. Try send something like this. ```json { "model": "gpt-5.4", "stream": false, "messages": [ { "role": "system", "content": "Say Hello to the user" }, { "role": "user", "content": "Hello" } ] } ``` 2. Then you will always get this: ``` { "error": { "message": "litellm.APIConnectionError: APIConnectionError: ChatgptException - Unknown items in responses API response: []. Received Model Group=gpt-5.4\nAvailable Model Group Fallbacks=None", "type": null, "param": null, "code": "500" } } ```

### Relevant log output

```shell litellm-1 | raise e litellm-1 | File "/app/.venv/lib/python3.13/site-packages/litellm/utils.py", line 1892, in wrapper_async litellm-1 | result = await original_function(*args, **kwargs) litellm-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ litellm-1 | File "/app/.venv/lib/python3.13/site-packag…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d18h48m15s ago — entered · #import:https:::github.com:berriai:litellm post #2059
A new provider integration carries broader scope and integration risk: provider routing, authentication, request/response translation, streaming and usage handling, tests, and documentation. The other issue is comparatively localized to diagnosing and correcting one non-streaming response path.
#0 of 0 · 31d18h19m46s ago — current · #import:https:::github.com:berriai:litellm post #2513
The left requires coordinated lifecycle handling across persistence synchronization, in-memory registries, updates, deletions, restarts, and multiple model-discovery paths, with broader regression and concurrency risk. The right is more likely an isolated provider-response parsing or non-streaming compatibility fix with narrower testing scope.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search