An IIT Kharagpur student recently tested OpenAI’s ChatGPT o3 model on the JEE Advanced 2025 paper, and the results were nothing short of stunning.
The AI scored an impressive 327 out of 360, a performance that would secure it All India Rank 4 in the real exam.
This score demonstrates ChatGPT o3’s advanced reasoning and problem-solving capabilities, especially in a notoriously challenging exam like JEE Advanced, known for testing deep conceptual understanding.
However, many are curious about the 33 marks it failed to secure.
Critics and educators alike are examining whether the errors were due to misinterpretation, lack of contextual understanding, or just simple mistakes — issues that still differentiate AI from top-performing human candidates.
While ChatGPT's near-perfect score showcases the future potential of AI in education and testing, it also sparks debates about the limits of machine intelligence when it comes to the very toughest questions.
