Is is acceptable to use ChatGPT to replace BERT etc. to do classfication tasks in papers?

Hi all! I used to use pretrained models like roberta to do some sentiment analysis and NER in my project. However, having tried gpt3.5 API, I found the results are better than my fine-tuned transformer models. Can I use ChatGPT to do those NLP tasks in my paper that aims to publish?

analysis api bert chatgpt etc found gpt3 gpt3.5 languagetechnology ner nlp pretrained models project roberta sentiment sentiment analysis tasks transformer transformer models