How Fast Do Agents Rot? An Empirical Study of Long-Horizon Degradation in LLM Agents for Production Decision-Making

arXiv:2609.01660 · physics.soc-ph, cs.AI · Submitted 2026-08-31 · Read on arXiv

physics.soc-ph, cs.AI

Submitted: 2026-08-31

Updated: 2026-08-31

Code: https://github.com/shubmittal/agent-horizon-degradation

Terminology

• Benchmark — 29× in this paper · explained in 2 episodes
• Accuracy — 6× in this paper · explained in 1 episode
• Agentic Task — 6× in this paper · explained in 1 episode
• Benchmarking — 4× in this paper · explained in 3 episodes
• Transformer — 2× in this paper · explained in 2 episodes
• AI Agents — 1× in this paper · explained in 2 episodes
• Agentic Evaluation — 1× in this paper · explained in 1 episode
• Agentic Loop — 1× in this paper · explained in 1 episode
• Agentic Workflows — 1× in this paper · explained in 1 episode
• Consistency — 1× in this paper · explained in 3 episodes
• Hallucination — 1× in this paper · explained in 3 episodes
• Large Language Model (LLM) — 1× in this paper · explained in 3 episodes
• Multi-Agent System — 1× in this paper · explained in 5 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Verifiability — 1× in this paper · explained in 2 episodes
• Verification — 1× in this paper · explained in 2 episodes
• Verifier — 1× in this paper · explained in 2 episodes

Sources

• ReAct: Synergizing Reasoning and Acting in Language Models
• Reflexion: Language Agents with Verbal Reinforcement Learning
• Faith and Fate: Limits of Transformers on Compositionality
• GAIA: a benchmark for General AI Assistants
• SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
• $\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
• WebArena: A Realistic Web Environment for Building Autonomous Agents
• Measuring AI Ability to Complete Long Software Tasks
• AI Agents That Matter
• LLMs Get Lost In Multi-Turn Conversation
• Mind2Web: Towards a Generalist Agent for the Web
• RULER: What's the Real Context Size of Your Long-Context Language Models?
• A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
• Same Task, More Tokens: the Impact of Input Length on the Reasoning Performance of Large Language Models
• AgentBench: Evaluating LLMs as Agents
• AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents
• OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments
• TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks

Related papers

• URBAN-SPIN: A street-level bikeability index to inform design implementations in historical city centres
• Civilizational Metamaterials: Engineering Coordination Under Capability Gradients and Structural Turbulence
• Collective Behavior of AI Agents: the Case of Moltbook
• A Distinct Communication Strategies Model of the Double Empathy Problem
• Measuring Primitive Accumulation: An Information-Theoretic Approach to Capitalist Enclosure in PIK2, Indonesia
• Diffusion-induced instabilities promote cooperation in eco-evolutionary networks