---
title: "Discussion on \"Qwen3.8 27B VRAM, measured: Q4_K_M uses 19.95 GiB at 32K context\" | Hashnode"
description: "Discussion on \"Qwen3.8 27B VRAM, measured: Q4_K_M uses 19.95 GiB at 32K context\". Real llama.cpp numbers, why the KV cache is only 64 KiB per token, and which GGUF fits a 16, 24 or 32 GB GPU"
url: "https://hashnode.com/posts/qwen3-8-27b-vram-measured-q4-k-m-uses-19-95-gib-at-32k-context/6accd6017fa5210e386410e5"
---

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

Discussion

# [Qwen3.8 27B VRAM, measured: Q4\_K\_M uses 19.95 GiB at 32K context](https://modelvram.hashnode.dev/qwen3-8-27b-vram-measured-q4-k-m-uses-19-95-gib-at-32k-context?utm_source=hashnode&utm_medium=feed)

Real llama.cpp numbers, why the KV cache is only 64 KiB per token, and which GGUF fits a 16, 24 or 32 GB GPU

Oct 12 · 5 min read · modelvram.hashnode.dev  
Read the article

-   [#llm](https://hashnode.com/tag/llm)
-   [#gpu](https://hashnode.com/tag/gpu)
-   [#llamacpp](https://hashnode.com/tag/llamacpp)
-   [#machine-learning](https://hashnode.com/tag/machine-learning)
-   [#ai](https://hashnode.com/tag/ai)

## Written by

[donald991234](https://hashnode.com/@donald991234)

@donald991234

  
[View profile](https://hashnode.com/@donald991234)

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
