🤖 ChatGPT vs the 2025 Intermediate Maths Challenge
🤖 The 2025 Intermediate Maths Challenge paper was released on Friday* and you can ​find it here​. As soon as I’d solved it myself, I decided to test ChatGPT on the paper too and the results were fascinating. I spent a lot of this weekend making this video as a result – not just testing the standard GPT-4 model, but also the latest o1 model that’s supposedly better at mathematical reasoning.
How well did it do? Well, you might be surprised at just how close AI is getting to matching the top students in this age group… check out its performance in the video below!
*Note: Due to the timings of the Chinese New Year meaning many students could not sit the IMC on the 29th, a different paper was used in Asia a few days before. I do not have any special info about certificate boundaries so my best guess is that they will be similar to previous years – I’ve compiled a table of these ​here​ and will not reply to e-mails asking for predictions!