QW

Huihui Qwen3 Next 80B A3B Thinking Abliterated

QwenGGUF by mradermacherUpdated 2025-12-22
Download on Hugging Face
Parameters80B
Context256K
Layers48
Architectureqwen3next
Smallest quant0.5 GB
Quick answer

Huihui Qwen3 Next 80B A3B Thinking Abliterated needs about 49.7 GB of memory at Q4_K_M (a 48.4 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 24 quants, from 0.5 GB to 65.5 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.

Huihui Qwen3 Next 80B A3B Thinking Abliterated – a large reasoning‑focused model

Huihui Qwen3 Next 80B A3B Thinking Abliterated is an 80 billion‑parameter Qwen3Next model with 3 billion active parameters in a mixture of experts. It was trained on 262,144 context tokens and is released under the MIT license. The model is available in multiple GGUF quantizations.

Good at

  • Advanced reasoning
  • Large context handling
  • Mixture of experts support

Limits

  • Requires large VRAM for full precision
  • No vision capabilities
  • No language metadata provided

Best for

  • Complex problem solving
  • Long‑form content generation
  • Research prototyping

Who it’s for. Developers and researchers needing a high‑capacity reasoning model for local inference. The model is suitable for environments with sufficient GPU memory and a need for large context windows.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.5 GB 1.8 GB
Fits Get file →
IQ1_S16.2 GB 17.5 GB
Too big Get file →
IQ1_M18.0 GB 19.3 GB
Too big Get file →
IQ2_XXS21.0 GB 22.3 GB
Too big Get file →
IQ2_XS23.4 GB 24.7 GB
Too big Get file →
IQ2_S23.6 GB 24.9 GB
Too big Get file →
IQ2_M26.1 GB 27.4 GB
Too big Get file →
Q2_K_S27.3 GB 28.6 GB
Too big Get file →
Q2_K29.1 GB 30.4 GB
Too big Get file →
IQ3_XXS30.8 GB 32.1 GB
Too big Get file →
IQ3_XS32.7 GB 34.0 GB
Too big Get file →
Q3_K_S34.5 GB 35.9 GB
Too big Get file →
IQ3_S34.5 GB 35.9 GB
Too big Get file →
IQ3_M35.0 GB 36.3 GB
Too big Get file →
Q3_K_M38.2 GB 39.5 GB
Too big Get file →
Q3_K_L41.2 GB 42.5 GB
Too big Get file →
IQ4_XS42.6 GB 43.9 GB
Too big Get file →
Q4_045.3 GB 46.6 GB
Too big Get file →
Q4_K_S45.5 GB 46.8 GB
Too big Get file →
Q4_K_M48.4 GB 49.7 GB
Too big Get file →
Q4_150.0 GB 51.3 GB
Too big Get file →
Q5_K_S55.0 GB 56.3 GB
Too big Get file →
Q5_K_M56.7 GB 58.0 GB
Too big Get file →
Q6_K65.5 GB 66.8 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (17.5 GB memory) or Q2_K_S (28.6 GB). For 16 GB VRAM, IQ2_S (24.7 GB) or Q3_K_S (35.9 GB) are appropriate. For 24 GB VRAM, IQ3_S (35.9 GB) or Q4_K_S (46.8 GB) fit the budget. Memory estimates are approximate.

What GPU do you need for Huihui Qwen3 Next 80B A3B Thinking 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: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
  • 6 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
  • 8 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
  • 12 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
  • 16 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
  • 24 GB VRAM: the best quant that fits is IQ2_XXS (21.0 GB file, about 22.3 GB in use).
  • 32 GB VRAM: the best quant that fits is Q2_K (29.1 GB file, about 30.4 GB in use).
  • 48 GB VRAM: the best quant that fits is Q4_K_S (45.5 GB file, about 46.8 GB in use).

Which Huihui Qwen3 Next 80B A3B Thinking Abliterated quant should you choose?

Q4_K_M is the usual starting point for Huihui Qwen3 Next 80B A3B Thinking Abliterated: it balances file size and quality well. Here it is a 48.4 GB file.

If memory is tight, the smallest file is the first one in the table at 0.5 GB, with the quality trade-offs that come with fewer bits.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen3next
Parameters80B (approximate, mixture of experts)
Active parameters3B
Layers48
Trained context262,144 tokens
KV cache (fp16)96 KB per token

Source

Quantized bymradermacher
FamilyQwen
Licensemit
Downloads (HF)5,580
Last updated2025-12-22

Frequently asked questions about Huihui Qwen3 Next 80B A3B Thinking Abliterated

How much VRAM does Huihui Qwen3 Next 80B A3B Thinking Abliterated need?

About 49.7 GB at Q4_K_M with an 8K context window. That is the 48.4 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 Next 80B A3B Thinking Abliterated on an 8 GB GPU?

Yes, with the smaller files: the 0.5 GB file needs about 1.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Huihui Qwen3 Next 80B A3B Thinking Abliterated GGUF?

It depends on the quant: file is 0.5 GB, IQ1_S is 16.2 GB, IQ1_M is 18.0 GB, IQ2_XXS is 21.0 GB, IQ2_XS is 23.4 GB, IQ2_S is 23.6 GB and 18 more.

What context length does Huihui Qwen3 Next 80B A3B Thinking Abliterated support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.

Where can I download Huihui Qwen3 Next 80B A3B Thinking Abliterated GGUF?

The files are on Hugging Face in the mradermacher/Huihui-Qwen3-Next-80B-A3B-Thinking-abliterated-i1-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 Next 80B A3B Thinking Abliterated use?

Hugging Face lists the license as mit. 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?

No, the smallest quant requires at least 17.5 GB of VRAM. Use a larger GPU or a smaller quant if available.

Does the model support vision tasks?

No, the model has no vision capabilities. It is text‑only.

What license applies to this model?

The model is released under the MIT license, allowing free use and modification.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Qwen models

All Qwen →

Other large models

All large models →