⌦abliteratedmodels.org

Qwen3.8-Flash-Next-DERISKED

Qwen3.8-Flash-Next-DERISKED is a abliterated and uncensored variant of Blackfrost-AI/Qwen3.8-Flash-Next-DERISKED-BF16, published on Hugging Face by davetha. It is other licence. It is distributed across 2 repositories in AWQ format, totalling 27.9K downloads.

Abliteration + uncensored fine-tuneQwen↓ 27,883♥ 4
View on Hugging Face ↗Base model ↗

Specification

Model name
Qwen3.8-Flash-Next-DERISKED
Source repository
davetha/Qwen3.8-Flash-Next-DERISKED-W4A16-AWQ
Base model
Blackfrost-AI/Qwen3.8-Flash-Next-DERISKED-BF16
Publisher
davetha
Parameters
Not determinable from name
VRAM required
Unknown — no published parameter count
Hardware class
Unknown
Licence
other
Task
Image text to text
Context window
256K tokens (as documented)
Ablation scope
Not stated by the publisher
Formats available
AWQ
First indexed
30 Aug 2026
Last updated
30 Aug 2026

Ablation technique

How Qwen3.8-Flash-Next-DERISKED was modified, and what that implies.

Abliteration + uncensored fine-tune

Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone.

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

“Built to run the 80B-class Flash-Next MoE on 2× AMD MI210 (gfx90a) at long context.”

Running it

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

Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "davetha/Qwen3.8-Flash-Next-DERISKED-W4A16-AWQ"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve davetha/Qwen3.8-Flash-Next-DERISKED-W4A16-AWQ --trust-remote-code

Downloads and variants (1)

The only published repository of Qwen3.8-Flash-Next-DERISKED.

RepositoryFormatDownloads
davetha/Qwen3.8-Flash-Next-DERISKED-W4A16-AWQsourceAWQ27.9K

Security-team assessment

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

  • Qwen3.8-Flash-Next-DERISKED 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.
  • The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
  • This release adds fine-tuning on unfiltered data, so its behaviour diverges from the base model beyond refusal removal — differences you measure cannot be attributed to ablation alone.
  • Declared licence is other, but the terms that bind you are Blackfrost-AI/Qwen3.8-Flash-Next-DERISKED-BF16’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 Qwen3.8-Flash-Next-DERISKED is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with davetha, and does not independently verify publisher claims. See responsible use.