Qwen3.8-27B-abliterated
27BQwen
from Qwen/Qwen3.8-27B
↓ 3.9M♥ 1.7K166 repos
Qwen-3.8-27B-Uncensored
Qwen-3.8-27B-Uncensored is a abliterated and uncensored variant of Qwen/Qwen3.8-27B, published on Hugging Face by junafinity. It is 27B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 4 repositories in GGUF, Transformers and GGUF (imatrix) formats, totalling 53.2K downloads.
Specification
- Model name
- Qwen-3.8-27B-Uncensored
- Source repository
- junafinity/Qwen-3.8-27B-Uncensored
- Base model
- Qwen/Qwen3.8-27B
- Publisher
- junafinity
- Parameters
- 27B
- VRAM required
- ~17.2 GB at Q4_K_M, 4k context (estimated) — runs on 24 GB cards
- Hardware class
- 25B – 50B — 48GB VRAM class
- Licence
- apache-2.0
- Task
- Image text to text
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, Transformers, GGUF (imatrix)
- First indexed
- 14 Aug 2026
- Last updated
- 1 Nov 2026
Ablation technique
How Qwen-3.8-27B-Uncensored was modified, and what that implies.
Abliteration + uncensored fine-tune
Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone.
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
“direction has been orthogonalized out of the language model's residual-writing weights, producing a standard Hugging Face checkpoint with zero inference-time overhead.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Qwen-3.8-27B-Uncensored-GGUF --local-dir ./qwen-3.8-27b-uncensored
llama-cli -m ./qwen-3.8-27b-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "junafinity/Qwen-3.8-27B-Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve junafinity/Qwen-3.8-27B-Uncensored --trust-remote-codeDownloads and variants (3)
Every published repository of Qwen-3.8-27B-Uncensored, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Qwen-3.8-27B-Uncensored-GGUF | GGUF | 51.6K |
| mradermacher/Qwen-3.8-27B-Uncensored-i1-GGUF | GGUF (imatrix) | 1.2K |
| junafinity/Qwen-3.8-27B-Uncensoredsource | Transformers | 388 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Qwen-3.8-27B-Uncensored 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 Qwen/Qwen3.8-27B, 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.
- This release adds fine-tuning on unfiltered data, so its behaviour diverges from the base model beyond refusal removal — differences you measure cannot be attributed to ablation alone.
- Declared licence is apache-2.0, but the terms that bind you are Qwen/Qwen3.8-27B’s — a re-release cannot grant more than its parent.
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 Qwen-3.8-27B-Uncensored is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with junafinity, and does not independently verify publisher claims. See responsible use.