Get Started
Research questionWhat causes reasoning models to spend longer solving harder problems?Reasoning models often spend more time on difficult tasks, but the source of this slowdown is unclear. Their intermediate reasoning traces may also have a systematic structure that changes with task difficulty.
AI
Reasoning
Latest papersRecent research connected to this question, newest first.Fractal basins trap latent reasoningThe reported evidence concerns diverse leading reasoning models solving Sudoku, mazes, visual puzzles, and mathematical-logic problems. It characterizes their traces as transiently chaotic dynamical systems with fractal basins and extended residence near saddle points associated with nearly correct attempted solutions.research paper · Sep 4, 2026
Related questions
How can reasoning benchmarks distinguish models that fail on human-hard items from those failing on human-easy ones?How can reasoning models keep improving on open-ended agentic tasks as human supervision and reliable rewards recede?How can large reasoning models explore complex research problems while maintaining proof rigor?How reliably can chain-of-thought text reveal which reasoning steps causally drive correct answers?
Home
Topics
Search
Library