Отправить ответ : Tencent improves testing originative AI models with unpractical benchmark