---
title: "How do DeepSeek API reasoning tokens count toward the output limit?"  
description: "How do DeepSeek API reasoning tokens count toward the output limit?"  
author: "Khoa Vu"  
published: 2026-10-22  
updated: 2026-10-22  
canonical: https://www.mindstick.com/forum/162755/how-do-deepseek-api-reasoning-tokens-count-toward-the-output-limit  
category: "DeepSeek API"  
tags: ["DeepSeek", "api(s)", "reasoning models", "Tokens"]  
reading_time: 1 minute  

---

# How do DeepSeek API reasoning tokens count toward the output limit?

I’m trying to plan output limits for requests to the DeepSeek API, especially when using a reasoning model. Do the tokens used for internal reasoning count toward the same output cap as the final answer, or are they handled separately?

If a response gets cut short, I’d like to know whether the reasoning portion can use up the available budget before the user-facing answer is complete. Any examples of how people size the limit for longer responses would help.


---

Original Source: https://www.mindstick.com/forum/162755/how-do-deepseek-api-reasoning-tokens-count-toward-the-output-limit

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