mlsanityData to 14 Oct 2026

How Consistent Are LLM Agents? Measuring Behavioral Reproducibility in Multi-Step Tool-Calling Pipelines

arXiv 2605.28840 ↗First posted 23 Apr 2026cs.CL cs.AI cs.SEcited by 5

On the map: LLM agents and tool use → Tool learning and function calling

Abstract (arXiv)

Large language model (LLM) agents with tool-calling capabilities are increasingly deployed in production systems, yet a fundamental reliability question remains under-explored: does the same agent behave the same way twice? We present a systematic empirical study of behavioral consistency in multi-step tool-calling agents, measuring whether agents select the same tools, in the same order, with the same arguments, across repeated identical invocations. Unlike prior work on consistency in ReAct-style agents(search-only, free-text actions), we study the richer setting of structured tool-calling interfaces with typed parameters and consequential side effects.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use. Builds on 13 of its 21 references that are in the map. First posted on arXiv 23 Apr 2026.

Builds on

13 of 21 references.

Closest ancestors · 4

Same direction, newest first

  1. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs31 Jul
  2. 2023Gorilla: Large Language Model Connected with Massive APIs24 May
  3. 2023API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs14 Apr
  4. 2023Toolformer: Language Models Can Teach Themselves to Use Tools9 Feb

Other references · 9

  1. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,926
  4. 2023AgentBench: Evaluating LLMs as Agentscited by 1,267
  5. 2021Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivitycited by 1,135
  6. 2021Ethical and social risks of harm from Language Modelscited by 966
  7. 2023Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formattingcited by 817
  8. 2021True Few-Shot Learning with Language Modelscited by 344
  9. 2024AI Agents That Mattercited by 219

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Determinants and Limits of LLM Security-Tool Orchestration: A Study with HexStrike-AIcited by 0
  2. 2026OpenCodeReview: Determinism over Non-Determinism for Cost-Effective Agent-Based Code Reviewcited by 0
  3. 2026Harness Engineering for Predictable Agentic Systems: An Empirical Study of Deterministic Execution Constraintscited by 0
  4. 2026Same Patient, Different Order: Action-Level Reliability of Clinical LLM Agents Under Repeated Runscited by 0
  5. 2026Prior or Feedback? What an LLM Uses When Adapting Neural Operatorscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026The Evolution of Tool Use in LLM Agents: From Single-Tool Call to Multi-Tool Orchestration
  2. 2026AgentJudgeBench: A Multi-Difficulty Benchmark for Evaluating LLM Judges on Agentic Tool-Calling
  3. 2026Verified Tool Calls Improve LLM Agent Reliability Under Non-Atomic Failures
  4. 2026ToolRobustBench: Stage-Wise Perturbation Evaluation and Failure Diagnosis for Tool-Calling Agents
  5. 2026Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias
  6. 2026Measuring Cross-Task Behavioral Consistency in Language Model Agents
  7. 2025Tool Preferences in Agentic LLMs are Unreliable
  8. 2026MANTRA: Synthesizing SMT-Validated Compliance Benchmarks for Tool-Using LLM Agents