How Consistent Are LLM Agents? Measuring Behavioral Reproducibility in Multi-Step Tool-Calling Pipelines
For developers deploying LLM agents in production, this work highlights a critical but overlooked reliability issue—behavioral inconsistency—that can lead to unpredictable side effects.
This paper systematically measures behavioral reproducibility in multi-step LLM agents with structured tool-calling interfaces, finding that even with temperature=0, agents exhibit non-determinism in tool selection, ordering, and arguments across repeated identical invocations.
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.