QW

Huihui Qwen3.8 27B Quark MXFP4 MagicQuant

QwenGGUF by gearwave00001Updated 2026-09-25
Download on Hugging Face
Parameters27.3B
Context256K
Layers65
Architectureqwen35
Smallest quant15.0 GB
Quick answer

Huihui Qwen3.8 27B Quark MXFP4 MagicQuant needs about 17.7 GB of memory at MXFP4 (a 15.0 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 2 quants, from 15.0 GB (MXFP4) to 18.9 GB (MXFP4). Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.

Huihui Qwen3.8 27B Quark MXFP4 MagicQuant – 27.3B parameter LLM with vision

The model is a 27.3B parameter Qwen variant, converted to AMD Quark MXFP4 format. It supports vision and has a trained context of 262144 tokens. The model is released under the apache-2.0 license.

Good at

  • Vision tasks
  • Large context handling
  • Low‑precision inference

Limits

  • No safety filtering – risk of sensitive content
  • Requires 8‑k context for memory estimates
  • Only MXFP4 quant available

Best for

  • Research on vision‑enabled LLMs
  • Testing low‑precision inference
  • Exploring large‑context generation

Who it’s for. Researchers and developers who need a 27B parameter Qwen model with vision support and are comfortable handling reduced safety filtering. The model is suitable for controlled environments and experimentation.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
MXFP415.0 GB 17.7 GB
Too big Get file →
MXFP418.9 GB 21.6 GB
Too big Get file →

For 8‑GB VRAM, use the 14.98 GB MXFP4 file; it estimates 17.7 GB memory at 8‑k context. For 12‑GB VRAM, use the 18.89 GB MXFP4 file; it estimates 21.6 GB memory at 8‑k context. Memory estimates are approximate.

What GPU do you need for Huihui Qwen3.8 27B Quark MXFP4 MagicQuant?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: the best quant that fits is MXFP4 (18.9 GB file, about 21.6 GB in use).

Which Huihui Qwen3.8 27B Quark MXFP4 MagicQuant quant should you choose?

MXFP4 (15.0 GB) is the middle option in this repository. Lower bit counts mean smaller files and more quality loss.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen35
Parameters27.3B
Layers65
Trained context262,144 tokens
KV cache (fp16)260 KB per token

Source

Quantized bygearwave00001
FamilyQwen
Licenseapache-2.0
Downloads (HF)569
Last updated2026-09-25

Frequently asked questions about Huihui Qwen3.8 27B Quark MXFP4 MagicQuant

How much VRAM does Huihui Qwen3.8 27B Quark MXFP4 MagicQuant need?

About 17.7 GB at MXFP4 with an 8K context window. That is the 15.0 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Huihui Qwen3.8 27B Quark MXFP4 MagicQuant on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 17.7 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Huihui Qwen3.8 27B Quark MXFP4 MagicQuant GGUF?

It depends on the quant: MXFP4 is 15.0 GB, MXFP4 is 18.9 GB.

What context length does Huihui Qwen3.8 27B Quark MXFP4 MagicQuant support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.2 GB at 8K and 8.7 GB at 32K.

Where can I download Huihui Qwen3.8 27B Quark MXFP4 MagicQuant GGUF?

The files are on Hugging Face in the gearwave00001/Huihui-Qwen3.8-27B-Quark-MXFP4-MagicQuant-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Huihui Qwen3.8 27B Quark MXFP4 MagicQuant use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run this model on 8 GB VRAM?

The 14.98 GB MXFP4 file estimates 17.7 GB memory at 8‑k context, so 8 GB VRAM is insufficient.

Does this model support text generation only?

No, the model has vision capabilities as detected by the repository.

Is the model safe for public use?

The model has reduced safety filtering; it is recommended for research or controlled environments only.

Other GGUFs of Huihui-Qwen3.8-27B-abliterated

Similar models

Browse all →

More Qwen models

All Qwen →

Other medium models

All medium models →