ai-agents-model-parameter-comp-when-using-openai-gpt-5-series-models-through-lite-02b98bdb

THEN Update requests for GPT-5 models to use 'max_completion_tokens' instead of 'max_tokens', set 'temperature' to 1 (the only supported value), and omit the 'reasoning' parameter entirely. Verify model-specific API documentation for correct parameter names and allowed values.

Corrected curl request:
```
curl --request POST \
  --url http://localhost:4000/v1/chat/completions \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "openai/gpt-5-mini",
  "messages": [{"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "hello in spanish?"}],
  "stream": false,
  "stream_options": {"include_usage": true},
  "max_completion_tokens": 2000,
  "temperature": 1
}'
```
Python snippet (paraphrased for license):
```python
response = litellm.completion(
    model="openai/gpt-5",
    messages=prompt,
    max_completion_tokens=100,
    temperature=1
)
```