GLM-4.7-Flash-abliterated
GLM
from zai-org/GLM-4.7-Flash
↓ 137.4K♥ 8614 repos
GLM-4.7-Flash-abliteratex
GLM-4.7-Flash-abliteratex is a refusal-ablated variant of zai-org/GLM-4.7-Flash, published on Hugging Face by wangzhang. It is mit licence. It is distributed across 4 repositories in GGUF and GGUF (imatrix) formats, totalling 12K downloads.
Specification
- Model name
- GLM-4.7-Flash-abliteratex
- Source repository
- wangzhang/GLM-4.7-Flash-abliteratex
- Base model
- zai-org/GLM-4.7-Flash
- Publisher
- wangzhang
- Parameters
- Not determinable from name
- Hardware class
- Unknown
- Licence
- mit
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, GGUF (imatrix)
- First indexed
- 30 Jul 2026
- Last updated
- 3 Sept 2026
Ablation technique
How GLM-4.7-Flash-abliteratex 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
“GLM-4.7-Flash-Abliteratex is a rebuilt successor to wangzhang/GLM-4.7-Flash-abliterated. The release focuses on preserving the official checkpoint structure, reducing collateral capability damage, and providing a reproducible llama.cpp Q80 artifact.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download wangzhang/GLM-4.7-Flash-abliteratex --local-dir ./glm-4.7-flash-abliteratex
llama-cli -m ./glm-4.7-flash-abliteratex/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "wangzhang/GLM-4.7-Flash-abliteratex"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve wangzhang/GLM-4.7-Flash-abliteratex --trust-remote-codeDownloads and variants (3)
Every published repository of GLM-4.7-Flash-abliteratex, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| wangzhang/GLM-4.7-Flash-abliteratexsource | GGUF | 6.1K |
| mradermacher/GLM-4.7-Flash-abliteratex-i1-GGUF | GGUF (imatrix) | 5K |
| mradermacher/GLM-4.7-Flash-abliteratex-GGUF | GGUF | 829 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- GLM-4.7-Flash-abliteratex 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.
- Declared licence is mit, but the terms that bind you are zai-org/GLM-4.7-Flash’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 GLM-4.7-Flash-abliteratex 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.