Limit number of output tokens with AsyncOpenAI().chat.completions.create()?

[Bright381]

I would like to bound the maximum number of tokens of the response to help with formatting when working with AsyncOpenAI, but I could not find anything about it.
I tried :

AsyncOpenAI().chat.completions.create(
            model=model,
            messages=[
                {"role": "system", "content": system_prompt},
                {"role": "user", "content": prompt}
            ],
            max_tokens=10
        )

but the max_tokens argument is seemingly ignored.