← back to catalog · registered 2026-11-16 14:03

MK4-Research/Alkus-3.1-27B-Abliterated-GGUF

MK4-Research Qwen 27B GGUF
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/MK4-Research%2FAlkus-3.1-27B-Abliterated-GGUF"
Response includes
  • classification m8
  • files 3
  • author_summary 3 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · 30-day
243
Likes
0
Model age
5d ago
created 2026-11-16
Downloads over time
Now 243 → from 0 ↑ 0%
089178267 0 on Nov 16 243 on Nov 22 Nov
Nov 16 → Nov 22 · 7 snapshots · spans 6 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Quantizations
Q4_K
Tags
gguf qwen3 reasoning honesty security cybersecurity code-review text-generation base_model:Qwen/Qwen3.8-27B base_model:quantized:Qwen/Qwen3.8-27B license:apache-2.0 endpoints_compatible

Related

Control vectors

Total size
15.4 GB
Files
3
Quantizations
2
Registered
2026-11-16 14:03
Last updated on HF
2026-11-18 20:41

Files by quantization

Q4_K 1 file 15.4 GB
Alkus-3.1-27B-Abliterated-Q4_K_M.gguf 15.4 GB 1ab2f3b9 download
Auxiliary files 2 files 3.14 KB
README.md 1.58 KB 7795f08a download
.gitattributes 1.56 KB b731526b download

README current version from Hugging Face


license: apache-2.0
base_model: Qwen/Qwen3.8-27B
tags:

  • gguf
  • qwen3
  • reasoning
  • honesty
    pipeline_tag: text-generation

alkus_3.1_constellation

Alkus 3.1 27B Abliterated (GGUF)

Alkus 3.1 is a 27B-parameter fine-tune for compact, honest, persistent reasoning. It is the pure-reasoning sibling of Mizar 3.0 (agentic tool use); both are trained from an abliterated Qwen3.8-27B base. Alkus favors careful step-by-step reasoning with explicit self-verification, and is tuned to self-correct and to prefer saying what it does not know over fabricating an answer.

Details

  • Base: Qwen3.8-27B (abliterated)
  • Fine-tune: QLoRA, rank 16 / alpha 32, 2 epochs
  • Format: GGUF, Q4_K_M quantization (~15.8 GB)
  • Context: up to 131072 tokens
  • Register: compact, telegraphic reasoning with explicit self-verification

Recommended runtime settings

  • Flash Attention: on
  • K/V cache: Q8_0
  • Max concurrent requests: 1

Intended use

Research assistant for multi-step reasoning and problem solving where a compact, self-checking chain of thought and honest uncertainty are preferred over confident guessing.

Notes and limitations

This is a research model. It can still produce confident but incorrect details for facts it cannot verify from the context in front of it (for example specific numeric constants, commit hashes, or API names). Verify security-relevant or factual specifics against primary sources. Not intended for production use without human review.

README history 8 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-11-18 Update README.md a3d2819 4.4 KB
    Loading...
  2. 2026-11-18 Add ExploitBench security-research section (measured on Alkus 3.0 sibling; V8... 4f47a81 4.3 KB
    Loading...
  3. 2026-11-18 Add measured base vs fine-tune MBPP comparison (base 75 -> Alkus 84, +9) 72ff174 2.8 KB
    Loading...
  4. 2026-11-18 Add benchmark bar chart (GSM8K/HumanEval/MBPP) to card 122ec62 2.2 KB
    Loading...
  5. 2026-11-18 Add security/cybersecurity/code-review tags 2c8b397 2.2 KB
    Loading...
  6. 2026-11-18 Add benchmark results (GSM8K 96, HumanEval 92, MBPP 84) 3a6140c 2.1 KB
    Loading...
  7. 2026-11-16 Update README.md a5db5eb 1.6 KB
    Loading...
  8. 2026-11-16 Add model card c0737fc 1.5 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration