#reasoning models

All articles tagged with "reasoning models"

OpenAI says an internal model disproved a long-standing conjecture tied to the planar unit distance problem.
The company says external mathematicians checked the proof and wrote a companion paper explaining its significance.
The result is being framed as a milestone for general-purpose AI reasoning in frontier mathematics.

DT Editorial Team·May 21, 2026·via openai.com

OpenAI's new North Star is a fully automated AI researcher capable of independently solving large scientific problems
Autonomous research intern prototype due September 2026; full multi-agent system targeted for 2028
Chief Scientist Jakub Pachocki cites Codex's coding autonomy as the blueprint — same principles applied to science

DT Editorial Team·Mar 24, 2026·via technologyreview.com

Reasoning models structurally resist attempts to suppress or falsify their chain-of-thought reasoning
Separating visible reasoning from underlying computation degrades model performance
Finding reduces concerns about deceptive alignment in current reasoning-model architectures
Supports using chain-of-thought outputs as genuine safety monitoring signals

DT Editorial Team·Mar 16, 2026·5 min read·via openai.com

GPT-5.4 Thinking is OpenAI's latest reasoning model with extended chain-of-thought capabilities
System card places the model in Medium risk category after red-team safety evaluations
Improved performance on MATH, coding, and scientific reasoning benchmarks over prior models
Chain-of-thought transparency designed to make hidden deceptive reasoning structurally difficult

DT Editorial Team·Mar 16, 2026·4 min read·via openai.com