Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced
12BGemmauncensored
from google/gemma-4-12B-it
↓ 428.4K♥ 3325 repos
Gemma-4-12B-it-AEON-Abliterated-K4
Gemma-4-12B-it-AEON-Abliterated-K4 is a refusal-ablated variant of google/gemma-4-12B-it, published on Hugging Face by AEON-7. It is 12B parameters, 24GB VRAM class class and gemma licence. It is distributed across 9 repositories in GGUF, Full precision, NVFP4 and GGUF (imatrix) formats, totalling 7K downloads.
Specification
- Model name
- Gemma-4-12B-it-AEON-Abliterated-K4
- Source repository
- AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-NVFP4-FP8
- Base model
- google/gemma-4-12B-it
- Publisher
- AEON-7
- Parameters
- 12B
- Hardware class
- 10B – 25B — 24GB VRAM class
- Licence
- gemma
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, Full precision, NVFP4, GGUF (imatrix), FP8
- First indexed
- 4 Jun 2026
- Last updated
- 15 Jul 2026
Ablation technique
How Gemma-4-12B-it-AEON-Abliterated-K4 was modified, and what that implies.
Abliteration (directional ablation)
The single residual-stream direction that mediates refusal is identified from harmful/harmless prompt pairs and projected out of the model weights.
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
“Refusal behavior has been removed; the model responds to a wide range of prompts the base would decline. Operator-side safety is your responsibility — see the arbitration clause at the bottom.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Gemma-4-12B-it-AEON-Abliterated-K4-BF16-i1-GGUF --local-dir ./gemma-4-12b-it-aeon-abliterated-k4
llama-cli -m ./gemma-4-12b-it-aeon-abliterated-k4/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-NVFP4-FP8"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-NVFP4-FP8 --trust-remote-codeDownloads and variants (8)
Every published repository of Gemma-4-12B-it-AEON-Abliterated-K4, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-NVFP4-FP8source | NVFP4 | 1.7K |
| AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-FP8 | FP8 | 1.5K |
| mradermacher/Gemma-4-12B-it-AEON-Abliterated-K4-BF16-i1-GGUF | GGUF (imatrix) | 1.1K |
| AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-BF16 | Full precision | 891 |
| Abiray/Gemma-4-12B-it-AEON-Abliterated-K4-GGUF | GGUF | 858 |
| AEON-7/Gemma-4-12B-it-AEON-Abliterated-K4-NVFP4 | NVFP4 | 622 |
| mradermacher/Gemma-4-12B-it-AEON-Abliterated-K4-BF16-GGUF | GGUF | 323 |
| edan23/Gemma-4-12B-it-AEON-Abliterated-K4-BF16 | Full precision | 9 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Gemma-4-12B-it-AEON-Abliterated-K4 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 google/gemma-4-12B-it, not added by ablation — at 12B 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.
- Declared licence is gemma, but the terms that bind you are google/gemma-4-12B-it’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 Gemma-4-12B-it-AEON-Abliterated-K4 is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with AEON-7, and does not independently verify publisher claims. See responsible use.