---
title: "Do DeepSeek function calls behave consistently across local inference engines?"  
description: "Do DeepSeek function calls behave consistently across local inference engines?"  
author: "Magdalena Malinowska"  
published: 2026-10-16  
updated: 2026-10-16  
canonical: https://www.mindstick.com/forum/162454/do-deepseek-function-calls-behave-consistently-across-local-inference-engines  
category: "DeepSeek Inference"  
tags: ["DeepSeek", "Local AI", "inference", "Function Calling"]  
reading_time: 1 minute  

---

# Do DeepSeek function calls behave consistently across local inference engines?

I’m testing DeepSeek models locally and getting different results from different inference engines. The model can produce a tool call in one setup, but in another it sometimes returns ordinary text that looks like a call instead.

For people who have compared local runtimes, how consistent is **DeepSeek function calling** across engines? Are the differences usually down to chat templates, parser support, or the specific model build?


---

Original Source: https://www.mindstick.com/forum/162454/do-deepseek-function-calls-behave-consistently-across-local-inference-engines

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