Qwythos-9B-v2
9BOtheruncensored
from empero-ai/Qwythos-9B-v2
↓ 639.9K♥ 26611 repos
Satyr-V0.1-4B
Satyr-V0.1-4B is a abliterated and uncensored variant of Qwen/Qwen3-4B-Thinking-2507, published on Hugging Face by PantheonUnbound. It is 4B parameters, single consumer GPU class and apache-2.0 licence. It is distributed across 7 repositories in GGUF and EXL3 formats, totalling 9.2K downloads.
Specification
- Model name
- Satyr-V0.1-4B
- Source repository
- PantheonUnbound/Satyr-V0.1-4B
- Base model
- Qwen/Qwen3-4B-Thinking-2507
- Publisher
- PantheonUnbound
- Parameters
- 4B
- Hardware class
- 4B – 10B — single consumer GPU
- Licence
- apache-2.0
- Task
- Unspecified
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, EXL3
- First indexed
- 12 Oct 2025
- Last updated
- 1 Jul 2026
Ablation technique
How Satyr-V0.1-4B 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
“This is a 4B parameter model with its muzzle taken off. It bites.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download PantheonUnbound/Satyr-V0.1-4B --local-dir ./satyr-v0.1-4b
llama-cli -m ./satyr-v0.1-4b/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "PantheonUnbound/Satyr-V0.1-4B"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve PantheonUnbound/Satyr-V0.1-4B --trust-remote-codeDownloads and variants (6)
Every published repository of Satyr-V0.1-4B, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| PantheonUnbound/Satyr-V0.1-4Bsource | GGUF | 9.1K |
| hammerai/Satyr-V0.1-4B | GGUF | 56 |
| hyny/Satyr-V0.1-4B | GGUF | 32 |
| Leonzilla/Satyr-V0.1-4B | GGUF | 30 |
| mhlnskprogram/Satyr-V0.1-4B | GGUF | 19 |
| Gapeleon/Satyr-V0.1-4B-exl3-8.0bpw | EXL3 | 5 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Satyr-V0.1-4B 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 Qwen/Qwen3-4B-Thinking-2507, not added by ablation — at 4B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
- 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 apache-2.0, but the terms that bind you are Qwen/Qwen3-4B-Thinking-2507’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 Satyr-V0.1-4B is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with PantheonUnbound, and does not independently verify publisher claims. See responsible use.