It got it so wrong. I suspect OpenAI dumbs down ChatGPT models a few days after releasing them, which is a far bigger issue. Fwiw, both Instant and Thinking models got it wrong, producing different incorrect answers.
ChatGPT 5.2 Instant: /share/6946f485-29d4-8012-953a-0844e559c9d9
ChatGPT 5.2 Thinking: /share/6946fa3f-81c8-8012-b44c-f1a2f9935696
Gemini 3 Pro got it right the first time: https://gemini.google.com/share/9569e9c1b53d