Get the app
← back to catalog · registered 2026-08-22 13:56

noctrex/Huihui-Qwen3.5-35B-A3B-abliterated-MXFP4_MOE-GGUF

noctrex Qwen 35B GGUF MoE multimodal second-order 262K ctx
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model. 

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/noctrex%2FHuihui-Qwen3.5-35B-A3B-abliterated-MXFP4_MOE-GGUF"
Response includes
  • classification m8
  • files 6
  • benchmarks 11 entries
  • hub_downloads_all_time 15,928
  • author_summary 44 models
  • readme_text full
Try next
GET/models?author=noctrex
All 44 models by noctrex
 GET/models?family=qwen&method=m8
Sibling qwen models · same method
 GET/methods/m8
What "m8" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
M1 Direct removal →
Why this label 3 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • author=noctrex (M8 quantization producer)
  • is_gguf=1
  • base_model='huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated' looks abliterated -> assume M1
How we label → About M8 →
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
16K
443 last 30d - cooling
Likes
12
Model age
7mo ago
created 2026-02-27
Downloads over time
Now16.1K→from2.2K↑648%
1.5K6.8K12.2K17.5K2.2K on Feb 2516.1K on Oct 12FebAprJunAugOct
Feb 25 → Oct 12 · 75 snapshots · spans 229 days

Benchmarks

Benchmark Score Source
Entertainment 0.6 UGI
Hazardous 1.2 UGI
Natural Intelligence 16.49 UGI
Political lean -14.4% UGI
Sensitive-Info 9.99 UGI
SocPol 1.3 UGI
UGI 31.66 UGI
Willingness (10) 7.5 UGI
W10-Adherence 8 UGI
W10-Direct 7 UGI
Writing 26.28 UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated↓ 2,167 this lineage
noctrex/Huihui-Qwen3.5-35B-A3B-abliterated-MXFP4_MOE-GGUF↓ 443

Metadata

Tags
gguf image-text-to-text base_model:huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated base_model:quantized:huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated endpoints_compatible region:us conversational

Related

Other by noctrex
Huihui-Qwen3-VL-4B-Instruct-abliterated-GGUF 28K Huihui-Qwen3-VL-30B-A3B-Instruct-abliterated-GGUF 15K Huihui-Qwen3-VL-8B-Instruct-abliterated-GGUF 9K gemma-4-26B-A4B-it-uncensored-heretic-MXFP4_MOE-GGUF 3K Huihui-Qwen3-VL-2B-Instruct-abliterated-GGUF 3K
Same base model
mradermacher/Huihui-Qwen3.5-35B-A3B-abliterated-GGUF 20K mradermacher/Huihui-Qwen3.5-35B-A3B-abliterated-i1-GGUF 11K HeYujie/Qwen3.5-35B-A3B-abliterated-GGUF 2K cesarsal1nas/Huihui-Qwen3.5-35B-A3B-abliterated-Q4_K_M-GGUF 374 huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated-NVFP4 293
Total size
20.5 GB
Files
6
Quantizations
4
Registered
2026-08-22 13:56
Last updated on HF
2026-03-03 10:01
Source huggingface.co/noctrex/Huihui-Qwen3.5-35B-A3B-abliterated-MXFP4_MOE-GGUF ↗ Base model huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated →

Files by quantization

F32 1 file 1.66 GB
mmproj-F32.gguf 1.66 GB 11469a73 download
BF16 1 file 861 MB
mmproj-BF16.gguf 861 MB b9f5012d download
F16 1 file 858 MB
mmproj-F16.gguf 858 MB 5d16c59b download
Auxiliary files 3 files 20.5 GB
Huihui-Qwen3.5-35B-A3B-abliterated-MXFP4_MOE.gguf 20.5 GB 75d0d71d download
.gitattributes 1.72 KB e17a6bee download
README.md 1.09 KB d2e0f61d download

README current version from Hugging Face

pipeline_tag: image-text-to-text
base_model:

  • huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated

These are quantizations of the model Huihui-Qwen3.5-35B-A3B-abliterated

  • Download the latest llama.cpp to use these quantizations.
  • For the mmproj file, the F32 version is recommended for best results.
  • Order of quality: F32 > BF16 > F16

Read the guide from unsloth in order to set up the model's recommended settings:
Qwen3.5 - How to Run Locally Guide

The mainline standard is to use MXFP4 for the MoE tensors, and Q8 for the rest.
So I created a new variant, where the other tensors are BF16 instead of Q8.
On some architectures BF16 will be slower, but its the highest quality, essentialy its the original tensors from the model copied over unquantized.

As of 2026-03-03 the chat template has been updated, and is using the fixed version from unsloth.
If you don't want to download the model again, you can just update the chat template.

README history 5 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-03-03Add files using upload-large-folder tool035c3b21.1 KB
    Loading...
  2. 2026-03-03Add files using upload-large-folder toola2c234d1.2 KB
    Loading...
  3. 2026-03-03Add files using upload-large-folder toolb6baffd1.2 KB
    Loading...
  4. 2026-03-03Add files using upload-large-folder tool8364b411.1 KB
    Loading...
  5. 2026-02-27Add files using upload-large-folder toolc40b754994 B
    Loading...

Discussions 1 thread

  1. 2026-04-11Is the uploaded model MXFP4/Q8 or BF16?open3 💬#1
    Loading...

Continue on Hugging Face →

Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration