OpenAI's GPT-5.4 solves unsolved math problem, earning praise from Fields Medal winner Terry Tao

OpenAI released GPT-5.4, an AI language model (software that understands and generates text) that solved a longstanding open problem in mathematics. The achievement was significant enough that renowned mathematician Terry Tao and researcher Jared Lichtman commented publicly on the result, indicating the math community is taking the breakthrough seriously.

Unlike earlier AI models used mainly for writing and customer service, GPT-5.4 demonstrated the ability to work through multi-step mathematical proofs (chains of logical reasoning) that humans had been unable to solve. This suggests the model can now handle abstract reasoning tasks that previously required specialized mathematical expertise.

For researchers and academics, this signals that AI is becoming a genuine research tool rather than just a productivity assistant—it can now contribute to solving open questions in mathematics, potentially accelerating discovery in fields like cryptography, optimization, and theoretical computer science that rely on mathematical breakthroughs.