---
title: "Discussion on \"Fixing CUDA PTX Error When Running Qwen3-VL with vLLM on H200\" | Hashnode"
description: "Discussion on \"Fixing CUDA PTX Error When Running Qwen3-VL with vLLM on H200\". Running vision-language models like Qwen3-VL with vLLM on high-end GPUs should be straightforward. Except when it's not.\n\nThe Problem\nI was setting up Qwen3-VL-8B-Instruct on our H200 cluster (8x H200, 143GB VRAM each) when I hit this error:\nvllm ser..."
url: "https://hashnode.com/posts/cuda-ptx-error-vllm-qwen3vl/69354b90de71fd22f3a21015"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Fixing CUDA PTX Error When Running Qwen3-VL with vLLM on H200](https://shaunliew.hashnode.dev/cuda-ptx-error-vllm-qwen3vl?utm_source=hashnode&utm_medium=feed)

Running vision-language models like Qwen3-VL with vLLM on high-end GPUs should be straightforward. Except when it's not. The Problem I was setting up Qwen3-VL-8B-Instruct on our H200 cluster (8x H200, 143GB VRAM each) when I hit this error: vllm ser...

Dec 7, 2025 · 4 min read · shaunliew.hashnode.dev  
Read the article

-   [#qwen3vl](https://hashnode.com/tag/qwen3vl)
-   [#vllm](https://hashnode.com/tag/vllm)

## Written by

[Shaun Liew](https://hashnode.com/@shaunliew20)

@shaunliew20 · Singapore

  
[View profile](https://hashnode.com/@shaunliew20)

computer vision, vlm, agentic ai

VLM Agentic AI Researcher

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
