QW

Qwen3.5 4B DFlash

QwenGGUF by magnitudedevUpdated 2026-10-07
Download on Hugging Face
Parameters634M
Context40K
Layers6
Architecturedflash
Smallest quant0.7 GB
Quick answer

Qwen3.5 4B DFlash needs about 1.4 GB of memory at Q8_0 (a 0.7 GB file) with an 8K context window, so it fits on a 4 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 0.8 GB.

Qwen3.5 4B DFlash – lightweight draft model for speculative decoding

Qwen3.5 4B DFlash is a draft model designed for DFlash speculative decoding. It is based on the Qwen family and uses a lightweight block diffusion draft approach. The model is intended to be paired with a target model in a speculative decoding server.

Good at

  • Fast draft proposals
  • Improved long‑context performance
  • Compatible with SGLang and vLLM

Limits

  • No standalone language model
  • Requires a target model for inference
  • No vision support

Best for

  • High‑throughput serving
  • Long‑context applications
  • Draft‑based inference pipelines

Who it’s for. Engineers building inference servers that support DFlash speculative decoding. The model is useful for those who need faster generation while maintaining the target model’s output distribution.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_00.7 GB 1.4 GB
Fits Get file →

The only available quant is Q8_0. The file size is 0.69 GB and requires approximately 1.4 GB of VRAM at 8 k context. This is suitable for systems with at least 8 GB of VRAM. No other quants are provided.

What GPU do you need for Qwen3.5 4B DFlash?

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 (0.7 GB file, about 1.4 GB in use).

Which Qwen3.5 4B DFlash quant should you choose?

Q8_0 (0.7 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
Parameters634M
Layers6
Trained context40,960 tokens
KV cache (fp16)24 KB per token

Source

Quantized bymagnitudedev
FamilyQwen
Licenseapache-2.0
Downloads (HF)3,698
Last updated2026-10-07

Frequently asked questions about Qwen3.5 4B DFlash

How much VRAM does Qwen3.5 4B DFlash need?

About 1.4 GB at Q8_0 with an 8K context window. That is the 0.7 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 Qwen3.5 4B DFlash on an 8 GB GPU?

Yes, with Q8_0: the 0.7 GB file needs about 1.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Qwen3.5 4B DFlash GGUF?

It depends on the quant: Q8_0 is 0.7 GB.

What context length does Qwen3.5 4B DFlash support?

The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.8 GB at 32K.

Where can I download Qwen3.5 4B DFlash GGUF?

The files are on Hugging Face in the magnitudedev/Qwen3.5-4B-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 Qwen3.5 4B DFlash 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 Qwen3.5 4B DFlash on 8 GB VRAM?

Yes, the Q8_0 quant requires about 1.4 GB of VRAM at 8 k context, which fits within an 8 GB GPU.

Does this model support vision tasks?

No, it is a language model draft and does not include vision capabilities.

What license applies to this model?

The model is released under the apache-2.0 license.

Other GGUFs of Qwen3.5-4B-DFlash

Similar models

Browse all →

More from magnitudedev

All by magnitudedev →

More Qwen models

All Qwen →

Other tiny models

All tiny models →