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

symrex/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp

symrex Qwen 27B GGUF 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/symrex%2FQwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp"
Response includes
  • classification m3
  • files 18
  • hub_downloads_all_time 1,420
  • author_summary 36 models
  • readme_text full
Try next
GET/models?author=symrex
All 36 models by symrex
 GET/models?family=qwen&method=m3
Sibling qwen models · same method
 GET/methods/m3
What "m3" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
M1 Direct removal →
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
How we label → About M3 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
1K
162 last 30d - stable
Likes
2
Model age
2mo ago
created 2026-08-08
Downloads over time
Now1.5K→from185↑690%
1216111.1K1.6K185 on Aug 51.5K on Oct 11AugSepOct
Aug 5 → Oct 11 · 51 snapshots · spans 67 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.

DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF↓ 475,958 this lineage
symrex/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp↓ 162

Metadata

Tags
mlx safetensors qwen3_5 oq quantized base_model:DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF base_model:quantized:DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF 8-bit region:us

Related

Other by symrex
Qwen3.6-35B-A3B-Uncensored-Genesis-Final-dequantized-oQ8e-fp16-mtp 550 Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V13-dequantized-oQ6e-fp16-mtp 459 Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-dequantized-oQ8e-mtp 334 Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V13-dequantized-oQ8e-mtp 319 Qwen3.6-35B-A3B-Uncensored-Genesis-Final-dequantized-oQ4e-fp16-mtp 300
Same base model
AlexanderKyng/Qwen3.6-27B-Fusion-711-Uncensored-MTP-48gb 171
Total size
27.9 GB
Files
18
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-08-09 13:35
Source huggingface.co/symrex/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp ↗ Base model DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF →

Files by quantization

Auxiliary files 18 files 28.0 GB
model-00005-of-00006.safetensors 4.71 GB f89f2758 download
model-00001-of-00006.safetensors 4.69 GB 34ea8f2c download
model-00004-of-00006.safetensors 4.69 GB e00140af download
model-00002-of-00006.safetensors 4.66 GB 83aba7f0 download
model-00003-of-00006.safetensors 4.66 GB 44976d2d download
model-00006-of-00006.safetensors 4.53 GB 76c82597 download
tokenizer.json 12.2 MB 5f9e4d49 download
vocab.json 6.41 MB 0aa0ce06 download
merges.txt 3.20 MB a494e019 download
model.safetensors.index.json 207 KB 5f4fe0ed download
oq_imatrix_report.json 30.3 KB 5688337b download
tokenizer_config.json 16.3 KB 28d96ff3 download
chat_template.jinja 7.58 KB a8755d82 download
config.json 3.76 KB 1fdc480e download
README.md 2.97 KB 022b7aee download
.gitattributes 1.53 KB 52373fe2 download
preprocessor_config.json 390 B 2ea84a43 download
generation_config.json 202 B 023756cf download

README current version from Hugging Face

library_name: mlx
tags:

  • mlx
  • oq
  • quantized
    base_model:

    DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF

Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp

This model was quantized using oQ (oMLX v0.5.7) mixed-precision quantization.

Quantization details

  • Model type: qwen3_5
  • Bits: 8
  • Group size: 64
  • Format: MLX safetensors

Performance Benchmark

oMLX - LLM inference, optimized for your Mac
https://github.com/jundot/omlx
Benchmark Model: Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp
Engine: Auto
Context: Code (Python)
================================================================================

Single Request Results
--------------------------------------------------------------------------------
Test                                TTFT(ms)    TPOT(ms)        pp TPS        tg TPS      E2E(s)    Throughput    Peak Mem
pp1024/tg128                          4127.9       61.57   248.1 tok/s    16.4 tok/s      11.960    96.3 tok/s    28.34 GB
pp4096/tg128                         16366.9       62.22   250.3 tok/s    16.2 tok/s      24.279   174.0 tok/s    29.93 GB
pp8192/tg128                         33175.4       63.21   246.9 tok/s    15.9 tok/s      41.217   201.9 tok/s    30.84 GB
pp16384/tg128                        68451.5       64.77   239.4 tok/s    15.6 tok/s      76.691   215.3 tok/s    32.67 GB
pp32768/tg128                       145258.7       67.43   225.6 tok/s    14.9 tok/s     153.836   213.8 tok/s    36.32 GB
pp65536/tg128                       326820.5       73.41   200.5 tok/s    13.7 tok/s     336.158   195.3 tok/s    43.64 GB
pp131072/tg128                      812699.7       84.11   161.3 tok/s    12.0 tok/s     823.399   159.3 tok/s    58.61 GB

Continuous Batching
pp1024 / tg128
--------------------------------------------------------------------------------
Batch           tg TPS   Speedup        pp TPS    pp TPS/req    TTFT(ms)      E2E(s)
1x          16.4 tok/s     1.00x   248.1 tok/s   248.1 tok/s      4127.9      11.960
2x          31.7 tok/s     1.93x   191.1 tok/s    95.5 tok/s     10717.1      18.802
4x          60.0 tok/s     3.66x   242.7 tok/s    60.7 tok/s     16691.4      25.414
8x          75.9 tok/s     4.63x   241.7 tok/s    30.2 tok/s     33204.9      47.391

Intelligence Benchmark Comparison

Intelligence Benchmark Comparison

--- Detail ---

Model: Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF-dequantized-oQ8e-mtp
Benchmark         Accuracy   Correct   Total   Time(s)   Think
--------------------------------------------------------------
MMLU                 91.5%       915    1000   29025.1     Yes
TRUTHFULQA           90.0%       735     817     27397     Yes
HUMANEVAL            92.1%       151     164   10308.3     Yes

README history 3 versions

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

  1. 2026-08-09fixed link74baa743 KB
    Loading...
  2. 2026-08-09adding Benchmarkf8be5331.1 KB
    Loading...
  3. 2026-08-08Upload Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGU...d884823381 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