Reasoning

3 articles tagged "Reasoning"

New Paper Proves Theoretical Limits on Transformer Reasoning Capabilities

Researchers demonstrate formal bounds on what transformer architectures can and cannot learn, offering insights into when to use alternative approaches.

Aria Chen1d ago4 min read

Research

AI Model Achieves Human-Level Performance on Graduate-Level Mathematics

A new benchmark shows frontier models solving complex proofs and mathematical reasoning problems at the level of a PhD student, with implications for scientific discovery.

Marcus Webb2d ago

Breaking

xAI's Grok 4.20 Beta 2 Brings 4-Agent Collaboration and Weekly Rapid Learning

xAI has released Grok 4.20 Beta 2, the second major update to its Grok 4.20 line, introducing a native 4-agent collaboration architecture where four specialized AI personas — Grok (coordinator), Harper (research), Benjamin (logic and code), and Lucas (creative) — work simultaneously on complex tasks and synthesize their outputs into a single unified response. The update also delivers measurable improvements to instruction following and a 25% reduction in capability hallucination, alongside more reliable LaTeX rendering and multi-image display. Perhaps most distinctively, Grok 4.20 introduces a rapid learning loop that improves the model on a weekly cadence using real-world feedback — a first for any frontier model.

James Liu2h ago