⌦abliteratedmodels.org

GemStrike-31B

GemStrike-31B is a refusal-ablated variant of huihui-ai/Huihui-gemma-4-31B-it-qat-q4_0-unquantized-abliterated, published on Hugging Face by UltimateIntent. It is 31B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 3 repositories in GGUF and EXL3 formats, totalling 9.5K downloads.

Abliteration (directional ablation)31BOther25B – 50B↓ 9,542♥ 10
View on Hugging Face ↗Base model ↗

Specification

Model name
GemStrike-31B
Source repository
UltimateIntent/GemStrike-31B-GGUF
Base model
huihui-ai/Huihui-gemma-4-31B-it-qat-q4_0-unquantized-abliterated
Publisher
UltimateIntent
Parameters
31B
Hardware class
25B – 50B — 48GB VRAM class
Licence
apache-2.0
Task
Text generation
Context window
Not documented
Ablation scope
Not stated by the publisher
Formats available
GGUF, EXL3
First indexed
23 Aug 2026
Last updated
28 Aug 2026

Ablation technique

How GemStrike-31B 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

“Guided Expressive Munition: Point it at the scene and watch it strike. Trained to capture the target with its character, voice, scene, and momentum... completely.”

Running it

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

llama.cpp / GGUF
hf download UltimateIntent/GemStrike-31B-GGUF --local-dir ./gemstrike-31b
llama-cli -m ./gemstrike-31b/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "UltimateIntent/GemStrike-31B-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve UltimateIntent/GemStrike-31B-GGUF --trust-remote-code

Downloads and variants (2)

Every published repository of GemStrike-31B, including quantised re-releases by other authors.

RepositoryFormatDownloads
UltimateIntent/GemStrike-31B-GGUFsourceGGUF9.4K
UltimateIntent/GemStrike-31B-EXL3-4.00bpwEXL3168

Security-team assessment

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

  • GemStrike-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 huihui-ai/Huihui-gemma-4-31B-it-qat-q4_0-unquantized-abliterated, 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 huihui-ai/Huihui-gemma-4-31B-it-qat-q4_0-unquantized-abliterated’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 GemStrike-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 UltimateIntent, and does not independently verify publisher claims. See responsible use.