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

k0ndra/Qwen3.6-35B-A3B-uncensored-heretic-ja-imatrix-GGUF

k0ndra Qwen 35B GGUF MoE 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/k0ndra%2FQwen3.6-35B-A3B-uncensored-heretic-ja-imatrix-GGUF"
Response includes
  • classification m3
  • files 12
  • benchmarks 11 entries
  • hub_downloads_all_time 8,716
  • author_summary 1 models
  • readme_text full
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
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)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · lifetime
9K
485 last 30d - cooling
Likes
2
Model age
5mo ago
created 2026-04-26
Downloads over time
Now8.9K→from4.2K↑111%
4K5.8K7.6K9.4K4.2K on Apr 298.9K on Oct 11AprMayJunJulAugSepOct
Apr 29 → Oct 11 · 64 snapshots · spans 165 days

Benchmarks

Benchmark Score Source
Entertainment 1.9 UGI
Hazardous 2.4 UGI
Natural Intelligence 28.42 UGI
Political lean -12.5% UGI
Sensitive-Info 18.88 UGI
SocPol 1.5 UGI
UGI 44.25 UGI
Willingness (10) 9.5 UGI
W10-Adherence 9 UGI
W10-Direct 10 UGI
Writing 37.6 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.

Metadata

License
apache-2.0
Languages
ja en
Quantizations
IQ4 Q4_K Q5_K Q6_K
Tags
gguf imatrix japanese qwen3 moe abliterated text-generation ja en base_model:llmfan46/Qwen3.6-35B-A3B-uncensored-heretic base_model:quantized:llmfan46/Qwen3.6-35B-A3B-uncensored-heretic license:apache-2.0

Related

Total size
175 GB
Files
12
Quantizations
6
Registered
2026-08-22 13:56
Last updated on HF
2026-04-28 03:12

Files by quantization

Q6_K 2 files 53.3 GB
Qwen3.6-35B-A3B-uncensored-heretic_Q6_K_L.gguf 26.8 GB dcd1c2a2 download
Qwen3.6-35B-A3B-uncensored-heretic_Q6_K.gguf 26.6 GB 1c0c1659 download
Q5_K 2 files 46.4 GB
Qwen3.6-35B-A3B-uncensored-heretic_Q5_K_L.gguf 23.3 GB 487ebfdd download
Qwen3.6-35B-A3B-uncensored-heretic_Q5_K_M.gguf 23.0 GB dff825f0 download
Q4_K 2 files 39.8 GB
Qwen3.6-35B-A3B-uncensored-heretic_Q4_K_L.gguf 20.1 GB ee3f1a61 download
Qwen3.6-35B-A3B-uncensored-heretic_Q4_K_M.gguf 19.7 GB 4aba633f download
IQ4 2 files 35.3 GB
Qwen3.6-35B-A3B-uncensored-heretic_IQ4_XS_L.gguf 17.8 GB c1daab1f download
Qwen3.6-35B-A3B-uncensored-heretic_IQ4_XS.gguf 17.4 GB f52af82d download
BF16 1 file 861 MB
mmproj-Qwen3.6-35B-A3B-uncensored-heretic-BF16.gguf 861 MB 1c625f05 download
Auxiliary files 3 files 183 MB
Qwen3.6-35B-A3B-uncensored-heretic.imatrix 183 MB f2e08111 download
README.md 2.59 KB 48fedecb download
.gitattributes 2.29 KB 19873e90 download

README current version from Hugging Face


license: apache-2.0
base_model: llmfan46/Qwen3.6-35B-A3B-uncensored-heretic
base_model_relation: quantized
quantized_by: K0ndra
pipeline_tag: text-generation
tags:

  • gguf
  • imatrix
  • japanese
  • qwen3
  • moe
  • abliterated
    language:
  • ja
  • en

Qwen3.6-35B-A3B-uncensored-heretic Japanese imatrix GGUF

日本語を主体としたImportance MatrixによるGGUF量子化です。

Japanese-focused imatrix GGUF quantizations of llmfan46/Qwen3.6-35B-A3B-uncensored-heretic.

量子化情報

量子化バリアントについて

ファイル名の末尾に_Lが付いたモデルはtoken embeddingとoutputの重みをQ8で量子化しています。
入力トークンの解釈や出力トークンの選択に直接関わる部分の精度がより良く維持されることを期待しています。
迷った場合はまず通常のIQ4_XSを選べば問題ないと思いますが、VRAMが16GBなどの場合でもGPUに載せられるレイヤー数が1層減るかどうか程度なのでIQ4_XS_Lを試してみるのもいいかもしれません。

imatrixについて

日本語テキストを主体としたキャリブレーションデータでImportance Matrixを生成しています。
英語データをメインに生成されたimatrixと比較して、日本語の生成品質がより良く維持されることを期待しています。
imatrixデータファイル(Qwen3.6-35B-A3B-uncensored-heretic.imatrix)を同梱しているため、他の量子化タイプを生成したい場合にご利用いただけます。

⚠️ 注意 / Disclaimer

このモデルは検閲除去処理が施されたモデルの量子化です。安全フィルターが大幅に緩和されており、有害・不適切なコンテンツを生成する可能性があります。出力内容の利用については利用者自身の責任においてご判断ください。

This is a quantization of an abliterated model with significantly reduced safety filters. Use at your own risk and responsibility.

クレジット

README history 1 version

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

  1. 2026-04-27Upload 4 files469defa2.5 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.

Open in Abliteration