---
title: Qwen3.8-27B UD-Q4_K_M on the Radeon RX 7900 XTX 24GB (Windows)
url: "https://bitworks.io/amd-inference/recipes/rx-7900-xtx-24gb-windows-qwen3-8-27b-q4km/"
description: Candidate Windows single-card configuration for RX 7900 XTX 24 GB; no published fit or performance result.
---

# Qwen3.8-27B UD-Q4_K_M on the Radeon RX 7900 XTX 24GB (Windows)

> Candidate Windows single-card configuration for RX 7900 XTX 24 GB; no published fit or performance result.

Human-readable page: https://bitworks.io/amd-inference/recipes/rx-7900-xtx-24gb-windows-qwen3-8-27b-q4km/

## Content

- **Status:** In lab testing (candidate)

- **GPU:** 1 × AMD Radeon RX 7900 XTX 24GB

- **Operating system:** Windows

- **Backend:** vulkan

- **Model:** Qwen3.8-27B UD-Q4_K_M — Unsloth GGUF conversion of Qwen/Qwen3.8-27B (Apache-2.0)

- **Context:** 32768 tokens

- **Concurrency:** 1

- **Placement:** full-gpu

No measured results are published for this recipe.

## Overview

This Windows candidate pairs the RX 7900 XTX 24 GB with the pinned model and quantization named above at the catalog's 32,768-token context. The 24 GB XTX candidate uses a different Q4 27B artifact and 32K catalog context than the 20 GB XT profile. Extra nominal memory permits a different proposed test, but does not prove the longer context or model placement succeeds. It is a test plan, not a claim of load, full-GPU placement, answer quality or speed.

## Requirements

Use a standard-user 64-bit Windows 11 lab system with an AMD driver exposing Vulkan and the Microsoft Visual C++ x64 runtime. The current control window supervises one loopback server; driver installation and a consumer installer are not included. Check the card's currently reported free GPU memory, not just the number printed on its box. A longer prompt, more active requests and the desktop can change the memory budget. Keep the exact driver and operating-system build with any later test.

The pinned GGUF is 16,464,440,224 bytes (16.46 GB decimal). Keep more free local SSD space than the artifact itself for verified acquisition and cache maintenance; no fixed extra margin or host-RAM minimum has been qualified. Review the exact model license before acquisition. A planned trial must record the actual free-memory budget, selected device, loaded context, host and GPU buffers, delivered output and representative answers. The catalog context is an intended setting, not a demonstrated maximum.

For agents: how to buy here — https://bitworks.io/checkout.md