GR

Huihui Granite 4.1 3b Abliterated

GraniteGGUF by huihui-aiUpdated 2026-05-03
Download on Hugging Face
Parameters3.4B
Context128K
Layers40
Architecturegranite
Smallest quant6.8 GB
Quick answer

Huihui Granite 4.1 3b Abliterated needs about 8.0 GB of memory at BF16 (a 6.8 GB file) with an 8K context window, so it fits on a 8 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 2.7 GB.

Huihui Granite 4.1 3b Abliterated – a 3.4B parameter instruction model

Huihui Granite 4.1 3b Abliterated is a 3.4B parameter model fine‑tuned from Granite‑4.1‑3B‑Base. It supports a 131072 token context window and is available in BF16 quantization. The model is released under Apache‑2.0 and is intended for instruction following and tool‑calling tasks.

Good at

  • Instruction following
  • Tool calling
  • Multilingual dialog

Limits

  • No vision support
  • Risk of sensitive or controversial outputs
  • Requires manual review

Best for

  • AI assistants
  • Tool‑enabled agents
  • Research testing

Who it’s for. Developers and researchers who need a large‑context instruction model with tool‑calling capabilities. The model is suitable for controlled environments and research use, not for production or public deployment.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
BF166.8 GB 8.0 GB
Tight Get file →

The BF16 quantization file is 6.81 GB and requires about 8 GB of VRAM at an 8 k context. For 8 GB VRAM, BF16 is the only available quant. For larger VRAM budgets, other quantizations could be considered if available.

What GPU do you need for Huihui Granite 4.1 3b 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: the best quant that fits is BF16 (6.8 GB file, about 8.0 GB in use).

Which Huihui Granite 4.1 3b Abliterated quant should you choose?

BF16 (6.8 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegranite
Parameters3.4B
Layers40
Trained context131,072 tokens
KV cache (fp16)80 KB per token

Source

Quantized byhuihui-ai
FamilyGranite
Licenseapache-2.0
Downloads (HF)2,515
Last updated2026-05-03

Frequently asked questions about Huihui Granite 4.1 3b Abliterated

How much VRAM does Huihui Granite 4.1 3b Abliterated need?

About 8.0 GB at BF16 with an 8K context window. That is the 6.8 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 Granite 4.1 3b Abliterated on an 8 GB GPU?

Yes, with BF16: the 6.8 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Huihui Granite 4.1 3b Abliterated GGUF?

It depends on the quant: BF16 is 6.8 GB.

What context length does Huihui Granite 4.1 3b Abliterated support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.7 GB at 32K.

Where can I download Huihui Granite 4.1 3b Abliterated GGUF?

The files are on Hugging Face in the huihui-ai/Huihui-granite-4.1-3b-abliterated 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 Granite 4.1 3b 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.

Can I run this model on 8 GB VRAM?

Yes, the BF16 quantization requires approximately 8 GB of VRAM at an 8 k context, which matches an 8 GB GPU.

Does this model support vision?

No, the model does not include vision capabilities.

What license does it use?

The model is released under the Apache‑2.0 license.

Other GGUFs of granite-4.1-3b

Similar models

Browse all →

More from huihui-ai

All by huihui-ai →

More Granite models

All Granite →

Other tiny models

All tiny models →