The paper comprehensively evaluates ChatGPT's performance on various academic tasks, covering 140 tasks across diverse fields, highlighting strengths and weaknesses, and introducing a new ability to follow multi-query instructions, ultimately paving the way for practical applications of ChatGPT-like models.