QW

Qwen3.8-27B-DFlash-v6-ep1

QwenGGUF by asf0Updated 2026-10-26
Download on Hugging Face
Parameters1.78B
Context256K
Layers5
Architecturedflash
Smallest quant1.9 GB
Quick answer

Qwen3.8-27B-DFlash-v6-ep1 needs about 2.6 GB of memory at Q8_0 (a 1.9 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 1.9 GB (Q8_0) to 3.6 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 0.7 GB.

Qwen3.8-27B-DFlash-v6-ep1 – a speculative‑decoding draft for the Qwen3.8‑27B verifier

Qwen3.8‑27B‑DFlash‑v6‑ep1 is a DFlash speculative‑decoding draft built from scratch against the Qwen3.8‑27B verifier. It is not a fine‑tune of the 27B weights and requires the verifier to function. The draft supports reasoning and tools abilities.

Good at

  • Speculative decoding with the Qwen3.8‑27B verifier
  • Reasoning tasks
  • Tool usage

Limits

  • Requires the Qwen3.8‑27B verifier
  • Not a standalone language model
  • Experimental research artifact

Best for

  • Speed‑up experiments with the verifier
  • Research on speculative decoding
  • Testing draft‑verifier pipelines

Who it’s for. Researchers and developers working on speculative decoding who have access to the Qwen3.8‑27B verifier. The model is intended for experimentation rather than production use.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_01.9 GB 2.6 GB
Fits Get file →
BF163.6 GB 4.2 GB
Fits Get file →

For systems with 8 GB VRAM, use the Q8_0 quantization (file size 1.91 GB, estimated 2.6 GB memory at 8 k context). If more memory is available, the BF16 quant (file size 3.58 GB, estimated 4.2 GB memory) offers higher precision. Both builds are approximate and should be tested for stability.

What GPU do you need for Qwen3.8-27B-DFlash-v6-ep1?

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: the best quant that fits is Q8_0 (1.9 GB file, about 2.6 GB in use).
  • 6 GB VRAM: the best quant that fits is BF16 (3.6 GB file, about 4.2 GB in use).

Which Qwen3.8-27B-DFlash-v6-ep1 quant should you choose?

Q8_0 (1.9 GB) is the middle option in this repository. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturedflash
Parameters1.78B
Layers5
Trained context262,144 tokens
KV cache (fp16)20 KB per token

Source

Quantized byasf0
FamilyQwen
Licenseapache-2.0
Downloads (HF)639
Last updated2026-10-26

Frequently asked questions about Qwen3.8-27B-DFlash-v6-ep1

How much VRAM does Qwen3.8-27B-DFlash-v6-ep1 need?

About 2.6 GB at Q8_0 with an 8K context window. That is the 1.9 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 Qwen3.8-27B-DFlash-v6-ep1 on an 8 GB GPU?

Yes, with BF16: the 3.6 GB file needs about 4.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Qwen3.8-27B-DFlash-v6-ep1 GGUF?

It depends on the quant: Q8_0 is 1.9 GB, BF16 is 3.6 GB.

What context length does Qwen3.8-27B-DFlash-v6-ep1 support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.7 GB at 32K.

Where can I download Qwen3.8-27B-DFlash-v6-ep1 GGUF?

The files are on Hugging Face in the asf0/Qwen3.8-27B-DFlash-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 Qwen3.8-27B-DFlash-v6-ep1 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 Qwen3.8‑27B‑DFlash‑v6‑ep1 on 8 GB VRAM?

Yes, the Q8_0 quantization requires about 2.6 GB memory at 8 k context, well within an 8 GB VRAM budget.

What license does this model use?

The model is released under the apache‑2.0 license.

Does this model work without the Qwen3.8‑27B verifier?

No, it is a draft that must be used together with the Qwen3.8‑27B verifier to function.

Similar models

Browse all →

More Qwen models

All Qwen →

Other tiny models

All tiny models →