A Survey of Frontiers in LLM Reasoning: Inference Scaling, Learning to Reason, and Agentic Systems

TL;DR

This survey categorizes and analyzes recent advancements in large language model reasoning, focusing on inference regimes, architectures, and emerging trends like learning-to-reason and agentic systems, to provide a systematic understanding of the field.

Contribution

It offers a comprehensive taxonomy of LLM reasoning methods, highlighting key trends and innovations in inference strategies, architectures, and workflows, with a focus on recent developments.

Findings

Shift from inference-scaling to learning-to-reason approaches

Transition towards agentic workflows incorporating external tools

Broad coverage of learning algorithms and reasoning architectures

Abstract

Reasoning is a fundamental cognitive process that enables logical inference, problem-solving, and decision-making. With the rapid advancement of large language models (LLMs), reasoning has emerged as a key capability that distinguishes advanced AI systems from conventional models that empower chatbots. In this survey, we categorize existing methods along two orthogonal dimensions: (1) Regimes, which define the stage at which reasoning is achieved (either at inference time or through dedicated training); and (2) Architectures, which determine the components involved in the reasoning process, distinguishing between standalone LLMs and agentic compound systems that incorporate external tools, and multi-agent collaborations. Within each dimension, we analyze two key perspectives: (1) Input level, which focuses on techniques that construct high-quality prompts that the LLM condition on; and (2) Output level, which methods that refine multiple sampled candidates to enhance reasoning quality. This categorization provides a systematic understanding of the evolving landscape of LLM reasoning, highlighting emerging trends such as the shift from inference-scaling to learning-to-reason (e.g., DeepSeek-R1), and the transition to agentic workflows (e.g., OpenAI Deep Research, Manus Agent). Additionally, we cover a broad spectrum of learning algorithms, from supervised fine-tuning to reinforcement learning such as PPO and GRPO, and the training of reasoners and verifiers. We also examine key designs of agentic workflows, from established patterns like generator-evaluator and LLM debate to recent innovations. ...