### Check for existing issues

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

### What happened?

Using the same Ollama model (`qwen3.5:4b`) and prompt locally, `litellm.completion()` returns the expected assistant text, but `litellm.responses()` returns a completed response object with empty text and no reasoning item.

### Steps to Reproduce

`uvx --with litellm python`

```py import litellm

resp1 = litellm.completion( model='ollama/qwen3.5:4b', messages=[{'role': 'user', 'content': 'Reply with exactly: ok'}], temperature=0, ) print('completion text', repr(resp1.choices[0].message.content))

resp2 = litellm.responses( model='ollama/qwen3.5:4b', input='Reply with exactly: ok', temperature=0, ) print('responses output:') print(resp2.output[0].model_dump_json(indent=2)) ```

Observed output:

``` completion text: 'ok' responses output: { "type": "message", "id": "chatcmpl-687cff74-119b-4158-9d76-99baf0e1eb56", "status": "completed", "role": "assistant", "content": [ { "type": "output_text", "text": "", "annotations": [] } ], "phase": null } ```

### What part of LiteLLM is this…