HU

Huihui MoE 12B A4B Abliterated

HuihuiGGUF by JohnRogerUpdated 2025-06-16
Download on Hugging Face
Parameters12.5B
Context40K
Layers36
Architectureqwen3moe
Smallest quant13.3 GB
Quick answer

Huihui MoE 12B A4B Abliterated needs about 15.0 GB of memory at Q8_0 (a 13.3 GB file) with an 8K context window, so it fits on a 16 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 4.8 GB.

Huihui MoE 12B A4B Abliterated is a 12.5B-parameter Huihui model in GGUF format, quantized by JohnRoger. Its trained context length is 40K tokens. The repository has 1 quant file, from 13.3 GB (Q8_0) to 13.3 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_013.3 GB 15.0 GB
Too big Get file →

What GPU do you need for Huihui MoE 12B A4B Abliterated?

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: the best quant that fits is Q8_0 (13.3 GB file, about 15.0 GB in use).

Which Huihui MoE 12B A4B Abliterated quant should you choose?

Q8_0 (13.3 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

Architectureqwen3moe
Parameters12.5B (approximate, mixture of experts)
Active parameters4B
Layers36
Trained context40,960 tokens
KV cache (fp16)144 KB per token

Source

Quantized byJohnRoger
FamilyHuihui
Licenseapache-2.0
Downloads (HF)24
Last updated2025-06-16

Frequently asked questions about Huihui MoE 12B A4B Abliterated

How much VRAM does Huihui MoE 12B A4B Abliterated need?

About 15.0 GB at Q8_0 with an 8K context window. That is the 13.3 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 MoE 12B A4B Abliterated on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 15.0 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Huihui MoE 12B A4B Abliterated GGUF?

It depends on the quant: Q8_0 is 13.3 GB.

What context length does Huihui MoE 12B A4B Abliterated support?

The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 1.2 GB at 8K and 4.8 GB at 32K.

Where can I download Huihui MoE 12B A4B Abliterated GGUF?

The files are on Hugging Face in the JohnRoger/Huihui-MoE-12B-A4B-abliterated-Q8_0-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 MoE 12B A4B Abliterated 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.

Other GGUFs of Huihui-MoE-12B-A4B-abliterated

Similar models

Browse all →

More from JohnRoger

All by JohnRoger →

More Huihui models

All Huihui →

Other medium models

All medium models →