# Non-Determinism of "Deterministic" LLM Settings

arXiv: https://arxiv.org/abs/2408.04667 · HTML: https://mlsanity.com/paper/2408.04667/

## Key facts

- Cited by 92 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 33 references that are in the map.
- First posted on arXiv 6 Aug 2024.

## Builds on (most cited first, 19 of 19)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark](https://mlsanity.com/paper/2406.01574/) (2024)
- [Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Them](https://mlsanity.com/paper/2210.09261/) (2022)
- [A Survey on Large Language Models for Code Generation](https://mlsanity.com/paper/2406.00515/) (2024)
- [Reasoning with Language Model Prompting: A Survey](https://mlsanity.com/paper/2212.09597/) (2022)
- [Leveraging Large Language Models for Multiple Choice Question Answering](https://mlsanity.com/paper/2210.12353/) (2022)
- [Are We Done with MMLU?](https://mlsanity.com/paper/2406.04127/) (2024)
- [Lessons from the Trenches on Reproducible Evaluation of Language Models](https://mlsanity.com/paper/2405.14782/) (2024)
- [An Empirical Study of the Non-determinism of ChatGPT in Code Generation](https://mlsanity.com/paper/2308.02828/) (2023)
- [Quantifying Uncertainty in Answers from any Language Model and Enhancing their Trustworthiness](https://mlsanity.com/paper/2308.16175/) (2023)
- [The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinism](https://mlsanity.com/paper/2407.10457/) (2024)
- [MultiLegalPile: A 689GB Multilingual Legal Corpus](https://mlsanity.com/paper/2306.02069/) (2023)
- [Cost-Effective Hyperparameter Optimization for Large Language Model Generation Inference](https://mlsanity.com/paper/2303.04673/) (2023)
- [The ART of LLM Refinement: Ask, Refine, and Trust](https://mlsanity.com/paper/2311.07961/) (2023)
- [Contextual Temperature for Language Modeling](https://mlsanity.com/paper/2012.13575/) (2020)
- [How Good Are LLMs at Out-of-Distribution Detection?](https://mlsanity.com/paper/2308.10261/) (2023)

## Built on by (most cited first, 30 of 92)

- [A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility](https://mlsanity.com/paper/2504.07086/) (2025)
- [RobustFlow: Towards Robust Agentic Workflow Generation](https://mlsanity.com/paper/2509.21834/) (2025)
- [On the Effect of Sampling Diversity in Scaling LLM Inference](https://mlsanity.com/paper/2502.11027/) (2025)
- [Evaluating Long-Context Reasoning in LLM-Based WebAgents](https://mlsanity.com/paper/2512.04307/) (2025)
- [Taming Uncertainty via Automation: Observing, Analyzing, and Optimizing Agentic AI Systems](https://mlsanity.com/paper/2507.11277/) (2025)
- [ACRFence: Preventing Semantic Rollback Attacks in Agent Checkpoint-Restore](https://mlsanity.com/paper/2603.20625/) (2026)
- [Evaluation and Facilitation of Online Discussions in the LLM Era: A Survey](https://mlsanity.com/paper/2503.01513/) (2025)
- [Challenges in Testing Large Language Model Based Software: A Faceted Taxonomy](https://mlsanity.com/paper/2503.00481/) (2025)
- [ATLAS: Adaptive Trading with LLM AgentS Through Dynamic Prompt Optimization and Multi-Agent Coordination](https://mlsanity.com/paper/2510.15949/) (2025)
- [LLM Output Drift: Cross-Provider Validation & Mitigation for Financial Workflows](https://mlsanity.com/paper/2511.07585/) (2025)
- [Toward Expert Investment Teams:A Multi-Agent LLM System with Fine-Grained Trading Tasks](https://mlsanity.com/paper/2602.23330/) (2026)
- [STED and Consistency Scoring: A Framework for Evaluating LLM Structured Output Reliability](https://mlsanity.com/paper/2512.23712/) (2025)
- [How Much Can Repeated Running Buy? A Measured Variance Ceiling for LLM Code Evaluation](https://mlsanity.com/paper/2606.00920/) (2026)
- [Through the Prism of Culture: Evaluating LLMs' Understanding of Indian Subcultures and Traditions](https://mlsanity.com/paper/2501.16748/) (2025)
- [MetaCipher: A Time-Persistent and Universal Multi-Agent Framework for Cipher-Based Jailbreak Attacks for LLMs](https://mlsanity.com/paper/2506.22557/) (2025)
- [Can LLMs Make (Personalized) Access Control Decisions?](https://mlsanity.com/paper/2511.20284/) (2025)
- [Beyond Reproducibility: Token Probabilities Expose Large Language Model Nondeterminism](https://mlsanity.com/paper/2601.06118/) (2026)
- [Compiled AI: Deterministic Code Generation for LLM-Based Workflow Automation](https://mlsanity.com/paper/2604.05150/) (2026)
- [The Non-Determinism of Small LLMs: Evidence of Low Answer Consistency in Repetition Trials of Standard Multiple-Choice Benchmarks](https://mlsanity.com/paper/2509.09705/) (2025)
- [Multi-Agent Taint Specification Extraction for Vulnerability Detection](https://mlsanity.com/paper/2601.10865/) (2026)
- [A Guide to Large Language Models in Modeling and Simulation: From Core Techniques to Critical Challenges](https://mlsanity.com/paper/2602.05883/) (2026)
- [Efficient Provably Secure Linguistic Steganography via Range Coding](https://mlsanity.com/paper/2604.08052/) (2026)
- [EQUITRIAGE: A Fairness Audit of Gender Bias in LLM-Based Emergency Department Triage](https://mlsanity.com/paper/2605.03998/) (2026)
- [FLIPS: Instance-Fingerprinting for LLMs via Pseudo-random Sequences](https://mlsanity.com/paper/2606.03330/) (2026)
- [Evaluating and Advancing Multimodal Large Language Models in Perception Ability Lens](https://mlsanity.com/paper/2411.14725/) (2024)
- [Designing Synthetic Discussion Generation Systems: A Case Study for Online Facilitation](https://mlsanity.com/paper/2503.16505/) (2025)
- [Something Just Like TRuST : Toxicity Recognition of Span and Target](https://mlsanity.com/paper/2506.02326/) (2025)
- [LLM Based Bayesian Optimization for Prompt Search](https://mlsanity.com/paper/2510.04384/) (2025)
- [ATA: A Neuro-Symbolic Approach to Implement Autonomous and Trustworthy Agents](https://mlsanity.com/paper/2510.16381/) (2025)
- [T-pro 2.0: An Efficient Russian Hybrid-Reasoning Model and Playground](https://mlsanity.com/paper/2512.10430/) (2025)

## Abstract (arXiv)

LLM (large language model) practitioners commonly notice that outputs can vary for the same inputs under settings expected to be deterministic. Yet the questions of how pervasive this is, and with what impact on results, have not to our knowledge been systematically investigated. We investigate non-determinism in five LLMs configured to be deterministic when applied to eight common tasks in across 10 runs, in both zero-shot and few-shot settings. We see accuracy variations up to 15% across naturally occurring runs with a gap of best possible performance to worst possible performance up to 70%. In fact, none of the LLMs consistently delivers repeatable accuracy across all tasks, much less identical output strings. Sharing preliminary results with insiders has revealed that non-determinism perhaps essential to the efficient use of compute resources via co-mingled data in input buffers so this issue is not going away anytime soon. To better quantify our observations, we introduce metrics focused on quantifying determinism, TARr@N for the total agreement rate at N runs over raw output, and TARa@N for total agreement rate of parsed-out answers. Our code and data are publicly available at https://github.com/breckbaldwin/llm-stability.