GPT-5 and Gemini 2.5 Pro won gold

Another medal for the AI collection: GPT-5 and Gemini 2.5 Pro won gold at the International Olympiad on Astronomy and Astrophysics (IOAA) To understand the difficulty of the Olympiad, keep in mind that it is not just about solving physics and math problems. Each task requires multi-step calculations, cross-checking with data and graphs, deep knowledge of astronomy and orbital mechanics, plotting charts on the celestial sphere, and more. But Gemini 2.5 Pro and GPT-5 showed almost perfect results – 85–86%. This is a gold medal and, moreover, a top-2 level among 200–300 participants in recent years. The weak spot, as expected, was geometric and spatial thinking + data analysis. So despite this result, AI still does not reach the level of even a weak researcher in this field. arxiv.org/abs/2510.05016