supergemma4-26b-uncensored-v2
26BOtheruncensored
from google/gemma-4-26B-A4B-it
↓ 394K♥ 98442 repos
Kanimus-26B-A4B-FFT-heretic
Kanimus-26B-A4B-FFT-heretic is a Heretic-ablated variant of Vortex5/Gemma-4-26B-A4B-Animus-V14.1-FFT-heretic, published on Hugging Face by SubMaroon. It is 26B total parameters (mixture-of-experts), 48GB VRAM class class and gemma licence. It is distributed across 4 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 8.4K downloads.
Specification
- Model name
- Kanimus-26B-A4B-FFT-heretic
- Source repository
- SubMaroon/Kanimus-26B-A4B-FFT-heretic
- Publisher
- SubMaroon
- Parameters
- 26B (mixture-of-experts)
- Hardware class
- 25B – 50B — 48GB VRAM class
- Licence
- gemma
- Task
- Image text to text
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix)
- First indexed
- 27 Aug 2026
- Last updated
- 31 Aug 2026
Ablation technique
How Kanimus-26B-A4B-FFT-heretic 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
“Experimental dark roleplay merge. Two edits on top of an abliterated Animus V14.1: a task-arithmetic delta injected into the attention Q and K projections from a Claude Opus distillation, and a stylistic LoRA baked in at low scale.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Kanimus-26B-A4B-FFT-heretic-i1-GGUF --local-dir ./kanimus-26b-a4b-fft-heretic
llama-cli -m ./kanimus-26b-a4b-fft-heretic/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "SubMaroon/Kanimus-26B-A4B-FFT-heretic"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve SubMaroon/Kanimus-26B-A4B-FFT-heretic --trust-remote-codeDownloads and variants (3)
Every published repository of Kanimus-26B-A4B-FFT-heretic, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Kanimus-26B-A4B-FFT-heretic-i1-GGUF | GGUF (imatrix) | 6.5K |
| mradermacher/Kanimus-26B-A4B-FFT-heretic-GGUF | GGUF | 1.7K |
| SubMaroon/Kanimus-26B-A4B-FFT-hereticsource | Transformers | 113 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Kanimus-26B-A4B-FFT-heretic 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 Vortex5/Gemma-4-26B-A4B-Animus-V14.1-FFT-heretic, not added by ablation — at 26B 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 Vortex5/Gemma-4-26B-A4B-Animus-V14.1-FFT-heretic’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 Kanimus-26B-A4B-FFT-heretic is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with SubMaroon, and does not independently verify publisher claims. See responsible use.