AI’s 48-Hour Frenzy: Breakthroughs, Billion-Dollar Bets & Bold Policies (July 22–23, 2025)
Google and OpenAI models each solved five of six problems at the International Mathematical Olympiad, earning gold medals with natural language-based reasoning. OpenAI scaled up test-time compute for its model, while Google DeepMind’s Gemini Deep Think completed the challenge in under 4.5 hours. MIT CSAIL and Facebook AI showed new image generation methods. Amazon closed its Shanghai AI lab amid U.S.–China tensions.