Qwen3.5-122B-A10B-abliterated
122BQwen
from wangzhang/Qwen3.5-122B-A10B-abliterix
↓ 26.1K♥ 3424 repos
Qwen3.5-122B-A10B-abliterix
Qwen3.5-122B-A10B-abliterix is a abliterated and uncensored variant of wangzhang/Qwen3.5-122B-A10B-abliterix, published on Hugging Face by wangzhang. It is 122B total parameters (mixture-of-experts), server-class deployment class and apache-2.0 licence. It is distributed across 7 repositories in Transformers, GGUF, GGUF (imatrix) and FP8 formats, totalling 20.2K downloads.
Specification
- Model name
- Qwen3.5-122B-A10B-abliterix
- Source repository
- wangzhang/Qwen3.5-122B-A10B-abliterix
- Base model
- wangzhang/Qwen3.5-122B-A10B-abliterix
- Publisher
- wangzhang
- Parameters
- 122B (mixture-of-experts)
- VRAM required
- ~74.7 GB at Q4_K_M, 4k context (estimated) — runs on 80 GB cards
- Hardware class
- 100B+ — server-class deployment
- Licence
- apache-2.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix), FP8, NVFP4
- First indexed
- 2 Mar 2026
- Last updated
- 2 Nov 2026
Ablation technique
How Qwen3.5-122B-A10B-abliterix 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.
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Qwen3.5-122B-A10B-abliterix-i1-GGUF --local-dir ./qwen3.5-122b-a10b-abliterix
llama-cli -m ./qwen3.5-122b-a10b-abliterix/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "wangzhang/Qwen3.5-122B-A10B-abliterix"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve wangzhang/Qwen3.5-122B-A10B-abliterix --trust-remote-codeDownloads and variants (6)
Every published repository of Qwen3.5-122B-A10B-abliterix, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| bjk110/Qwen3.5-122B-A10B-abliterix-NVFP4 | NVFP4 | 15.6K |
| mradermacher/Qwen3.5-122B-A10B-abliterix-i1-GGUF | GGUF (imatrix) | 4K |
| mradermacher/Qwen3.5-122B-A10B-abliterix-GGUF | GGUF | 431 |
| groxaxo/Qwen3.5-122B-A10B-abliterix-UD-Q4_K_XL-GGUF | GGUF | 60 |
| bjk110/Qwen3.5-122B-A10B-abliterix-FP8 | FP8 | 40 |
| wangzhang/Qwen3.5-122B-A10B-abliterixsource | Transformers | 33 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Qwen3.5-122B-A10B-abliterix 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 wangzhang/Qwen3.5-122B-A10B-abliterix, not added by ablation — at 122B 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.
- 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 wangzhang/Qwen3.5-122B-A10B-abliterix’s — a re-release cannot grant more than its parent.
- The source repository is gated, so downloading requires accepting the publisher’s terms.
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 Qwen3.5-122B-A10B-abliterix is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with wangzhang, and does not independently verify publisher claims. See responsible use.