OpenAI's GPT-5.5 Pro model has independently solved open problems in number theory, generating complete research preprints without human assistance. A notable mathematician described the output as being of solid doctoral quality. This development suggests a potential paradigm shift in scientific research, with AI taking on complex theoretical tasks. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Sets new SOTA on theoretical math problems; signals potential for AI to autonomously generate scientific research.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]