Preprint · 2026
How Consistent Are LLM Agents? Measuring Behavioral Reproducibility in Multi-Step Tool-Calling Pipelines
| Venue | arXiv |
|---|---|
| URL | https://arxiv.org/abs/2605.28840 |
| Topics | ReplicHAI / Agent & LLM run-to-run variability |
| Tags | replichai |
| archiveID | arXiv:2605.28840 |
| citationKey | yagubyan2026consistent |
| libraryCatalog | background/bib/bib-q1q2.bib |
BibTeX
@misc{yagubyan2026consistent,
title = {How {Consistent} {Are} {LLM} {Agents}? {Measuring} {Behavioral} {Reproducibility} in {Multi}-{Step} {Tool}-{Calling} {Pipelines}},
url = {https://arxiv.org/abs/2605.28840},
publisher = {arXiv},
author = {Yagubyan, Abel},
year = {2026},
keywords = {replichai},
}