Return to Article Details Reasoning capabilities of large language models: a systematic review and meta-analysis of benchmarks, methods, and emergent behaviours (2018–2025) Download Download PDF