The content was modified in parallel function calling

The content replied for parallel function calling was modified for unlnown reason. Here’s the case I met in Traditional chinese:

{
    'id': 'chatcmpl-8URutVoVauOIigsz36LLuwlB9Fgka',
    'object': 'chat.completion',
    'created': 1702267843,
    'model': 'gpt-4-1106-preview',
    'choices': [
        {
            'index': 0,
            'message': {'role': 'assistant', 'content': '2023年金馬獎影后是林品儒,而金曲獎歌后是A-Lin。'},
            'finish_reason': 'stop'
        }
    ],
    'usage': {'prompt_tokens': 2260, 'completion_tokens': 34, 'total_tokens': 2294},
    'system_fingerprint': 'fp_a24b4d720c'
}

The reply includes the name of an actress called ‘林品儒’ that came from the data retruned to parallel function calling. If it’s not the case of parallel function calling, the replied is

{
    'id': 'chatcmpl-8URtJkE1M2vQNZJfwZt2qDvrDYdnp',
    'object': 'chat.completion',
    'created': 1702267745,
    'model': 'gpt-4-1106-preview',
    'choices': [
        {
            'index': 0,
            'message': {
                'role': 'assistant',
                'content': '2023金馬獎影后獲獎人是林品彤,她憑藉在電影《小曉》中的表演獲此殊榮。'
            },
            'finish_reason': 'stop'
        }
    ],
    'usage': {'prompt_tokens': 569, 'completion_tokens': 56, 'total_tokens': 625},
    'system_fingerprint': 'fp_a24b4d720c'
}

The name of the actress is ‘林品彤’, the correct answer.

I’m not sure if this issue occurs in other languages, but it’s really bothersome.