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

AITRADER/Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated-mlx-mxfp8

AITRADER Qwen 27B multimodal second-order
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/AITRADER%2FHuihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated-mlx-mxfp8"
Response includes
  • classification m1
  • files 18
  • hub_downloads_all_time 2,985
  • author_summary 30 models
  • readme_text full
Try next
GET/models?author=AITRADER
All 30 models by AITRADER
 GET/models?family=qwen&method=m1
Sibling qwen models · same method
 GET/methods/m1
What "m1" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
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=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
How we label → About M1 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
3K
298 last 30d - cooling
Likes
5
Model age
6mo ago
created 2026-03-28
Downloads over time
Now3.1K→from384↑714%
2471.3K2.3K3.4K384 on Apr 13.1K on Oct 12AprMayJunJulAugSepOct
Apr 1 → Oct 12 · 68 snapshots · spans 194 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/Qwen3.5-27B-abliterated this lineage
AITRADER/Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated-mlx-mxfp8↓ 298

Metadata

License
apache-2.0
Tags
mlx safetensors qwen3_5 vision qwen3.5 abliterated tool-use function-calling mxfp8 image-text-to-text conversational license:apache-2.0

Related

Other by AITRADER
Jackrong-Qwen3.5-9B-Claude-Reasoning-abliterated-mxfp8-MLX 263 Huihui-Qwen3-Coder-Next-abliterated-mlx-8Bit 251 gemma-4-31B-Claude-Opus-Distill-abliterated-mlx-mxfp8 211 Qwen2.5-VL-32B-Instruct-abliterated-mlx-fp16 203 Huihui-Qwen3-Coder-Next-abliterated-mlx-4Bit 192
Same base model
AITRADER/Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated-mlx-8bit 68
Total size
26.7 GB
Files
18
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-03-28 22:00
Source huggingface.co/AITRADER/Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated-mlx-mxfp8 ↗ Base model huihui-ai/Qwen3.5-27B-abliterated ↗

Files by quantization

Auxiliary files 18 files 26.7 GB
model-00002-of-00007.safetensors 5.00 GB 4704d615 download
model-00004-of-00007.safetensors 5.00 GB 298ee4cd download
model-00001-of-00007.safetensors 4.99 GB 74a9fe26 download
model-00003-of-00007.safetensors 4.95 GB ae226e8e download
model-00005-of-00007.safetensors 4.67 GB eced1ed8 download
model-00006-of-00007.safetensors 1.22 GB 61ae5488 download
model-00007-of-00007.safetensors 879 MB 6bd134e2 download
tokenizer.json 19.1 MB 3019cb5b download
model.safetensors.index.json 157 KB a4868982 download
chat_template.jinja 7.57 KB a585dec8 download
config.json 3.87 KB 419488d1 download
.gitattributes 1.53 KB 52373fe2 download
README.md 1.38 KB abad7db2 download
tokenizer_config.json 1.28 KB a2d85d5e download
processor_config.json 1.27 KB 7ad6acdf download
preprocessor_config.json 337 B e501264c download
video_preprocessor_config.json 332 B 4d1c4049 download
generation_config.json 178 B bb6b4de8 download

README current version from Hugging Face

base_model: huihui-ai/Qwen3.5-27B-abliterated
tags:

  • mlx
  • vision
  • qwen3.5
  • abliterated
  • tool-use
  • function-calling
  • mxfp8
    license: apache-2.0
    pipeline_tag: image-text-to-text
    library_name: mlx

Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated MLX MXFP8

MXFP8 (Microscaling FP8) quantized MLX version of Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated.

Model Details

  • Architecture: Qwen 3.5 27B (hybrid linear attention + full attention)
  • Quantization: MXFP8 (E4M3 with block-level scaling), group_size=32
  • Size: ~29 GB
  • Context Length: 262,144 tokens
  • Vision: Full image and video understanding via integrated vision tower (27 ViT blocks, kept in bf16)
  • Tool Use: Native function calling support
  • Thinking: Chain-of-thought reasoning mode

Why MXFP8?

MXFP8 uses floating-point (E4M3) representation with per-block scaling instead of fixed-point integer quantization. This gives:

  • Better handling of outlier weights (exponent absorbs magnitude)
  • Lower quantization error across varying tensor ranges
  • Native hardware acceleration on modern chips

Capabilities

  • Image understanding and description
  • Video understanding
  • Tool use / function calling
  • Multi-step agent reasoning
  • Thinking/reasoning mode
  • Multilingual support
  • Long context (262K tokens)

Usage

Works with LM Studio, MLX, and other MLX-compatible frameworks.

README history 1 version

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

  1. 2026-03-28Add Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated MLX MXFP8 with vision49beefe1.4 KB
    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