GSM-Agent: Understanding Agentic Reasoning Using Controllable Environments

• URL: http://arxiv.org/abs/2509.21998v2
• Date: Thu, 02 Oct 2025 07:34:19 GMT
• Title: GSM-Agent: Understanding Agentic Reasoning Using Controllable Environments
• Authors: Hanlin Zhu, Tianyu Guo, Song Mei, Stuart Russell, Nikhil Ghosh, Alberto Bietti, Jiantao Jiao,
• Abstract summary: GSM-Agent is a novel benchmark for evaluating agentic reasoning in complex environments.<n>We analyze the agentic reasoning patterns by cluster the environment's document embeddings into nodes, and map each tool call to its nearest node.<n>We propose a tool-augmented test-time scaling method to improve LLM's agentic reasoning performance by adding tools to encourage models to revisit.
• Score: 56.007498767771075
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: As LLMs are increasingly deployed as agents, agentic reasoning - the ability to combine tool use, especially search, and reasoning - becomes a critical skill. However, it is hard to disentangle agentic reasoning when evaluated in complex environments and tasks. Current agent benchmarks often mix agentic reasoning with challenging math reasoning, expert-level knowledge, and other advanced capabilities. To fill this gap, we build a novel benchmark, GSM-Agent, where an LLM agent is required to solve grade-school-level reasoning problems, but is only presented with the question in the prompt without the premises that contain the necessary information to solve the task, and needs to proactively collect that information using tools. Although the original tasks are grade-school math problems, we observe that even frontier models like GPT-5 only achieve 67% accuracy. To understand and analyze the agentic reasoning patterns, we propose the concept of agentic reasoning graph: cluster the environment's document embeddings into nodes, and map each tool call to its nearest node to build a reasoning path. Surprisingly, we identify that the ability to revisit a previously visited node, widely taken as a crucial pattern in static reasoning, is often missing for agentic reasoning for many models. Based on the insight, we propose a tool-augmented test-time scaling method to improve LLM's agentic reasoning performance by adding tools to encourage models to revisit. We expect our benchmark and the agentic reasoning framework to aid future studies of understanding and pushing the boundaries of agentic reasoning.

Related papers

• Agent-Dice: Disentangling Knowledge Updates via Geometric Consensus for Agent Continual Learning [41.461840578204956]
  Large Language Model (LLM)-based agents learn new tasks without catastrophic forgetting.<n>Agent-Dice is a parameter fusion framework based on directional consensus evaluation.<n>Experiments on GUI agents and tool-use agent domains demonstrate that Agent-Dice exhibits outstanding continual learning performance.
  arXiv  Detail & Related papers  (2026-01-07T06:43:50Z)
• Multimodal Reinforcement Learning with Agentic Verifier for AI Agents [131.46008226323423]
  Argos is a principled multimodal reward agent to train reasoning models for agentic tasks.<n>By leveraging our agentic verifier across both SFT data and RL training, our model achieves state-of-the-art results.
  arXiv  Detail & Related papers  (2025-12-03T04:42:47Z)
• AgentPRM: Process Reward Models for LLM Agents via Step-Wise Promise and Progress [71.02263260394261]
  Large language models (LLMs) still encounter challenges in multi-turn decision-making tasks.<n>We build process reward models (PRMs) to evaluate each decision and guide the agent's decision-making process.<n>AgentPRM captures both the interdependence between sequential decisions and their contribution to the final goal.
  arXiv  Detail & Related papers  (2025-11-11T14:57:54Z)
• SFR-DeepResearch: Towards Effective Reinforcement Learning for Autonomously Reasoning Single Agents [93.26456498576181]
  This paper focuses on the development of native Autonomous Single-Agent models for Deep Research.<n>Our best variant SFR-DR-20B achieves up to 28.7% on Humanity's Last Exam benchmark.
  arXiv  Detail & Related papers  (2025-09-08T02:07:09Z)
• GDS Agent: A Graph Algorithmic Reasoning Agent [0.0]
  We introduce the GDS (Graph Data Science) agent in this technical report.<n>The GDS agent introduces a comprehensive set of graph algorithms as tools, together with preprocessing (retrieval) and postprocessing of algorithm results.<n>Results indicate that GDS agent is able to solve a wide spectrum of graph tasks.
  arXiv  Detail & Related papers  (2025-08-28T10:35:44Z)
• Distilling Tool Knowledge into Language Models via Back-Translated Traces [12.670632885715305]
  We propose a new paradigm for distilling tool knowledge into large language models (LLMs) purely through natural language.<n>A Translator Agent generates explanations for individual tool calls, while a Rephrase Agent merges them into a fluent and globally coherent narrative.<n>We show that fine-tuning a small open-source model on these synthesized traces enables it to internalize both tool knowledge and structured reasoning patterns.
  arXiv  Detail & Related papers  (2025-06-23T22:10:38Z)
• Agent-X: Evaluating Deep Multimodal Reasoning in Vision-Centric Agentic Tasks [94.19506319646376]
  We introduce Agent-X, a benchmark for evaluating vision-centric agents in real-world, multimodal settings.<n>Agent-X features 828 agentic tasks with authentic visual contexts, including images, multi-image comparisons, videos, and instructional text.<n>Our results reveal that even the best-performing models, including GPT, Gemini, and Qwen families, struggle to solve multi-step vision tasks.
  arXiv  Detail & Related papers  (2025-05-30T17:59:53Z)
• ThinkGeo: Evaluating Tool-Augmented Agents for Remote Sensing Tasks [54.52092001110694]
  ThinkGeo is a benchmark designed to evaluate tool-augmented agents on remote sensing tasks via structured tool use and multi-step planning.<n>Inspired by tool-interaction paradigms, ThinkGeo includes human-curated queries spanning a wide range of real-world applications.<n>Our analysis reveals notable disparities in tool accuracy and planning consistency across models.
  arXiv  Detail & Related papers  (2025-05-29T17:59:38Z)
• Distilling LLM Agent into Small Models with Retrieval and Code Tools [57.61747522001781]
  Agent Distillation is a framework for transferring reasoning capability and task-solving behavior from large language models into small language models.<n>Our results show that sLMs as small as 0.5B, 1.5B, 3B parameters can achieve performance competitive with next-tier larger 1.5B, 3B, 7B models.
  arXiv  Detail & Related papers  (2025-05-23T08:20:15Z)
• ReMA: Learning to Meta-think for LLMs with Multi-Agent Reinforcement Learning [53.817538122688944]
  We introduce Reinforced Meta-thinking Agents (ReMA) to elicit meta-thinking behaviors from Reasoning of Large Language Models (LLMs)<n>ReMA decouples the reasoning process into two hierarchical agents: a high-level meta-thinking agent responsible for generating strategic oversight and plans, and a low-level reasoning agent for detailed executions.<n> Empirical results from single-turn experiments demonstrate that ReMA outperforms single-agent RL baselines on complex reasoning tasks.
  arXiv  Detail & Related papers  (2025-03-12T16:05:31Z)
• Agentic Reasoning: A Streamlined Framework for Enhancing LLM Reasoning with Agentic Tools [19.70178343422698]
  We introduce Agentic Reasoning, a framework that enhances large language model (LLM) reasoning by integrating external tool-using agents.<n>Key innovation in our framework is the Mind-Map agent, which constructs a structured knowledge graph to store reasoning context.<n>When deployed on DeepSeek-R1, our method achieves a new state-of-the-art (SOTA) among public models.
  arXiv  Detail & Related papers  (2025-02-07T04:08:46Z)
• Improving LLM Reasoning with Multi-Agent Tree-of-Thought Validator Agent [9.439315294704368]
  Tree of Thoughts (ToT) methods have shown potential in improving reasoning for complex question-answering tasks. A critical limitation in multi-agent reasoning is the 'Reasoner' agent's shallow exploration of reasoning paths. We introduce a novel approach combining ToT-based Reasoner agents with a Thought Validator agent. Our method demonstrates superior performance compared to existing techniques when evaluated on the GSM8K dataset.
  arXiv  Detail & Related papers  (2024-09-17T19:54:37Z)
• GANterfactual-RL: Understanding Reinforcement Learning Agents' Strategies through Visual Counterfactual Explanations [0.7874708385247353]
  We propose a novel but simple method to generate counterfactual explanations for RL agents. Our method is fully model-agnostic and we demonstrate that it outperforms the only previous method in several computational metrics.
  arXiv  Detail & Related papers  (2023-02-24T15:29:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.