---
title: "How repeatable are DeepSeek’s answers with identical prompts?"  
description: "How repeatable are DeepSeek’s answers with identical prompts?"  
author: "Otto Lehtonen"  
published: 2026-10-15  
updated: 2026-10-15  
canonical: https://www.mindstick.com/forum/162311/how-repeatable-are-deepseek-s-answers-with-identical-prompts  
category: "LLM Evaluation"  
tags: ["DeepSeek", "LLM Evaluation", "AI Testing"]  
reading_time: 1 minute  

---

# How repeatable are DeepSeek’s answers with identical prompts?

I’m testing DeepSeek on a task where small differences between runs can make evaluation tricky. With the [same prompt](https://answers.mindstick.com/qa/117423/why-can-gpt-give-different-answers-to-the-same-prompt-on-different-runs) and settings, do you usually get substantially the same answer, or does the reasoning wander enough to affect the result?

I’d be interested in what you hold constant during a comparison—model version, sampling settings, or anything else—and whether you’ve found a practical way to distinguish normal variation from a real change in behavior.


---

Original Source: https://www.mindstick.com/forum/162311/how-repeatable-are-deepseek-s-answers-with-identical-prompts

Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.