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

elbelga/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated_MXFP4_MOE

elbelga 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/elbelga%2FHuihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated_MXFP4_MOE"
Response includes
  • classification m8
  • files 6
  • hub_downloads_all_time 4,164
  • author_summary 3 models
  • readme_text full
Try next
GET/models?author=elbelga
All 3 models by elbelga
 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
MEDIUM
Inherited from base model
M1 Direct removal →
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
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
4K
142 last 30d - cooling
Likes
2
Model age
6mo ago
created 2026-03-26
Downloads over time
Now4.2K→from302↑1,300%
1061.6K3.1K4.6K302 on Mar 254.2K on Oct 11MarAprMayJunJulAugSepOct
Mar 25 → Oct 11 · 68 snapshots · spans 200 days

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-Claude-4.6-Opus-abliterated↓ 256 this lineage
elbelga/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated_MXFP4_MOE↓ 142

Metadata

Tags
gguf image-text-to-text base_model:huihui-ai/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated base_model:quantized:huihui-ai/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated endpoints_compatible region:us imatrix conversational

Related

Other by elbelga
Huihui-Qwen3.6-35B-A3B-abliterated_MXFP4_MOE 231 Huihui-gemma-4-31B-it-abliterated-v2-MXFP4 29
Same base model
cesarsal1nas/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-Q4_K_M-GGUF 33K mradermacher/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-i1-GGUF 2K mradermacher/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-GGUF 826 lhca521/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-GGUF 633 tstepspam/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-Q8-MLX 424
Total size
20.5 GB
Files
6
Quantizations
4
Registered
2026-08-22 13:56
Last updated on HF
2026-03-26 20:25
Source huggingface.co/elbelga/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated_MXFP4_MOE ↗ Base model huihui-ai/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated →

Files by quantization

mmproj 1 file 1.66 GB
mmproj-f32.gguf 1.66 GB 329ba10e download
BF16 1 file 861 MB
mmproj-bf16.gguf 861 MB b46afc94 download
F16 1 file 858 MB
mmproj-f16.gguf 858 MB 3c67b55f download
Auxiliary files 3 files 20.5 GB
Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-MXFP4_MOE.gguf 20.5 GB 52e76a19 download
.gitattributes 1.74 KB b64fb474 download
README.md 944 B 56d2856d download

README current version from Hugging Face

base_model:

  • huihui-ai/Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated
    pipeline_tag: image-text-to-text

This is a quantization of the model Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated

Setup Instructions:

  • Download the latest llama.cpp.
  • Use the F32 version of the mmproj file for optimal results. Recommended quality ranking: F32 > BF16 > F16.

For configuration tips, follow the Unsloth Qwen3.5 local run guide

About this Variant:
The typical standard uses MXFP4 for MoE tensors and Q8 for everything else. In this build, I switched the non-MoE tensors to BF16. While this may result in slower inference on certain architectures, it offers the highest possible quality, effectively keeping the tensor values unquantized (or closest to the originals).

README history 4 versions

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

  1. 2026-03-26Update README.md7b04e8d944 B
    Loading...
  2. 2026-03-26Update README.mde6bf1ad828 B
    Loading...
  3. 2026-03-26Update README.md62b3c13946 B
    Loading...
  4. 2026-03-26Add files using upload-large-folder tool87debc6955 B
    Loading...
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