New test reveals how well AI models solve logic puzzles

Original: Satisfiability Solving with LLMs: A Matched-Pair Evaluation of Reasoning Capability

Source: arxiv.org ↗