# To connect Discourse to the DeepSeek Open Platform, the VLLM provider is required

**URL:** <https://meta.discourse.org/t/discourse-deepseek-vllm/415947>\
**Category:** General\
**Tags:** ai\
**Created:** [October 28, 2026, 1:50am UTC](https://meta.discourse.org/t/discourse-deepseek-vllm/415947 "2026-10-28T01:50:17Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![AkarinLiu](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/akarinliu/32/539577_2.png) [@AkarinLiu](https://meta.discourse.org/u/AkarinLiu)\
**Post date:** [October 28, 2026, 1:50am UTC](https://meta.discourse.org/t/discourse-deepseek-vllm/415947/1 "2026-10-28T01:50:17Z")

</div>

I found this solution while chatting with DeepSeek. Here are the steps:

1. Create an API Key on the [DeepSeek Open Platform](https://platform.deepseek.com);
2. Create an AI model in Discourse;
3. Select VLLM as the provider;
4. Set the reasoning parser to `deepseek_v4`;
5. Test and save.

I hope my method helps you. DeepSeek chat log: [https://chat.deepseek.com/share/egrylxjiubaxmm8rkh](https://chat.deepseek.com/share/egrylxjiubaxmm8rkh)