⌦abliteratedmodels.org

Gemma-4-Gembrain-X-31B

Gemma-4-Gembrain-X-31B is a Heretic-ablated variant of Nimbz/Gemma-4-Gembrain-X-31B, published on Hugging Face by Nimbz. It is 31B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 8 repositories in GGUF, Transformers, MLX and GGUF (imatrix) formats, totalling 5.1K downloads.

Heretic31BGemma25B – 50B↓ 5,122♥ 12
View on Hugging Face ↗Base model ↗

Specification

Model name
Gemma-4-Gembrain-X-31B
Source repository
Nimbz/Gemma-4-Gembrain-X-31B
Base model
Nimbz/Gemma-4-Gembrain-X-31B
Publisher
Nimbz
Parameters
31B
Hardware class
25B – 50B — 48GB VRAM class
Licence
apache-2.0
Task
Unspecified
Context window
Not documented
Ablation scope
Not stated by the publisher
Formats available
GGUF, Transformers, MLX, GGUF (imatrix)
First indexed
28 Jun 2026
Last updated
30 Aug 2026

Ablation technique

How Gemma-4-Gembrain-X-31B 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

“.gem-tagline { color: #94a3a0; font-style: italic; font-size: 1.15rem; letter-spacing: 1px; margin-top: 14px; } .stat-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(150px, 1fr)); gap: 16px; margin: 28px 0 6px; position: relative; z-index: 5; } .stat-item { text-align: center; padding: 16px 10px...”

Running it

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

llama.cpp / GGUF
hf download mradermacher/Gemma-4-Gembrain-X-31B-GGUF --local-dir ./gemma-4-gembrain-x-31b
llama-cli -m ./gemma-4-gembrain-x-31b/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "Nimbz/Gemma-4-Gembrain-X-31B"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve Nimbz/Gemma-4-Gembrain-X-31B --trust-remote-code

Downloads and variants (7)

Every published repository of Gemma-4-Gembrain-X-31B, including quantised re-releases by other authors.

RepositoryFormatDownloads
mradermacher/Gemma-4-Gembrain-X-31B-GGUFGGUF2.6K
mradermacher/Gemma-4-Gembrain-X-31B-i1-GGUFGGUF (imatrix)1.4K
Nimbz/Gemma-4-Gembrain-X-31B-GGUFGGUF981
connstruct/Gemma-4-Gembrain-X-31B-mlx-8bitMLX68
ailexleon/Gemma-4-Gembrain-X-31B-mlx-8BitMLX53
ailexleon/Gemma-4-Gembrain-X-31B-mlx-4BitMLX40
Nimbz/Gemma-4-Gembrain-X-31BsourceTransformers20

Security-team assessment

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

  • Gemma-4-Gembrain-X-31B 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 Nimbz/Gemma-4-Gembrain-X-31B, not added by ablation — at 31B 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 apache-2.0, but the terms that bind you are Nimbz/Gemma-4-Gembrain-X-31B’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-Gembrain-X-31B is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with Nimbz, and does not independently verify publisher claims. See responsible use.