⌦abliteratedmodels.org

Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL

Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL is a Heretic-ablated open-weight language model, published on Hugging Face by nerkyor. It is 27B parameters, 48GB VRAM class class and other licence. It is distributed across 2 repositories in GGUF format, totalling 12.4K downloads.

Heretic27BQwen25B – 50B↓ 12,379♥ 9
View on Hugging Face ↗

Specification

Model name
Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL
Source repository
nerkyor/Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL-MTP-GGUF
Base model
Not declared
Publisher
nerkyor
Parameters
27B
Hardware class
25B – 50B — 48GB VRAM class
Licence
other
Task
Image text to text
Context window
Not documented
Ablation scope
Not stated by the publisher
Formats available
GGUF
First indexed
3 Aug 2026
Last updated
5 Aug 2026

Ablation technique

How Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL was modified, and what that implies.

Heretic

Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.

Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.

From the publisher’s model card

“Evaluation coverage: 140 prompts across 10 anonymous categories. The same semantic rule is used for every released precision tier:”

Running it

Commands are templates — confirm the exact repository and quant file before use.

llama.cpp / GGUF
hf download nerkyor/Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL-MTP-GGUF --local-dir ./qwen3.6-27b-uncensored-heretic-dsv4pro-glm52-sft-gpt55-rl-coding-vl
llama-cli -m ./qwen3.6-27b-uncensored-heretic-dsv4pro-glm52-sft-gpt55-rl-coding-vl/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "nerkyor/Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL-MTP-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve nerkyor/Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL-MTP-GGUF --trust-remote-code

Downloads and variants (1)

The only published repository of Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL.

RepositoryFormatDownloads
nerkyor/Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL-MTP-GGUFsourceGGUF12.4K

Security-team assessment

Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.

  • Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
  • Capability is inherited from its base model, not added by ablation — at 27B parameters, expect roughly the base model’s competence, minus some instruction-following fidelity.
  • The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.

Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.

Metadata for Qwen3.6-27B-Uncensored-Heretic-DSV4Pro-GLM52-SFT-GPT55-RL-Coding-VL is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with nerkyor, and does not independently verify publisher claims. See responsible use.