OpenAI released a batch of AI-driven mathematical research results covering 372 result families and solving hundreds of long-standing open problems. The findings indicate that each result requires compute equivalent to three hours of continuous ChatGPT Pro reasoning, sparking academic debate on AI research ethics and paradigms.
Termsslug: logical-reasoning
Logical Reasoning
逻辑推理
The model's ability to derive conclusions from premises, often used in proofs
Curated Intelligence
IT之家 · 智能时代T2·Models
68 pts
OpenAI 官方博客T1·Models
85 pts
An OpenAI model solved the 80-year-old unit distance problem, disproving a major conjecture in discrete geometry. This breakthrough marks a milestone in AI-driven mathematics, demonstrating capabilities beyond simple pattern matching.
Google DeepMindT1·Models
85 pts
Google has officially launched Gemini 3, its most intelligent AI model, featuring significant enhancements in reasoning and multimodal capabilities. The model is now integrated across core products including the Gemini app, AI Studio, and Vertex AI, with a 'Deep Think' mode planned for Ultra subscribers soon. Additional models in the Gemini 3 series are expected to follow.