To investigate the potential of GPT-4.0 feedback to substitute teacher feedback in flipped classrooms, we conducted research that GPT-4.0 and a course teacher provided cognitive, praise, and mitigating feedback weekly via a course discussion platform, respectively. At the end of each topic, 89 third-year university students majoring in Educational Technology completed a self-regulated learning questionnaire, learning satisfaction questionnaire, and learning performance test without knowledge of the feedback provider's identity. The results indicate no significant difference between GPT-4.0 and teacher feedback in promoting students' self-regulated learning skills in the context of cognitive feedback. However, students generally expressed higher satisfaction with traditional teacher feedback, and those receiving teacher feedback outperformed those receiving GPT-4.0 feedback. The findings suggest that while GPT-4.0 provides cognitive feedback efficiently, its effectiveness in praising and mitigating feedback is limited compared to real teachers. This study reveals the potential and limitations of GPT-4.0 as a supplementary, rather than a substitute, teaching tool in the flipped classroom.