OpenAI's Astra Model Solves Ten Long-Standing Mathematical Problems

OpenAI has demonstrated an internal version of its upcoming model family, Astra, solving ten open problems in mathematics and theoretical computer science that had seen no progress for at least a decade. The company achieved these solutions at a cost of under $2,000 per problem, calculated at GPT-5.6 Sol token pricing. This achievement represents a major milestone in AI reasoning and automated theorem proving, signaling a shift toward "big mathematics" where AI handles complex technical tasks. It also intensifies the competition between OpenAI and Anthropic in solving highly complex, domain-specific scientific challenges. OpenAI released the Lean 4 formalizations of the proofs in a public GitHub repository, along with a paper and an LLM-generated PDF reconstructing the model's reasoning traces. However, the company did not disclose the exact prompts used or how many unsuccessful attempts were made.