GLM-4.7-Flash-abliterated
GLM
from zai-org/GLM-4.7-Flash
↓ 137.4K♥ 8614 repos
GLM-4.7-Flash-Derestricted
GLM-4.7-Flash-Derestricted is a abliterated and uncensored variant of koute/GLM-4.7-Flash-Derestricted, published on Hugging Face by koute. It is mit licence. It is distributed across 5 repositories in Transformers, GGUF, GGUF (imatrix) and MLX formats, totalling 132.3K downloads.
Specification
- Model name
- GLM-4.7-Flash-Derestricted
- Source repository
- koute/GLM-4.7-Flash-Derestricted
- Base model
- koute/GLM-4.7-Flash-Derestricted
- Publisher
- koute
- 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
- Transformers, GGUF, GGUF (imatrix), MLX
- First indexed
- 24 Jan 2026
- Last updated
- 26 Jan 2026
Ablation technique
How GLM-4.7-Flash-Derestricted 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 GLM-4.7-Flash model which has been uncensored using the Norm-Preserving Biprojected Abliteration methodology, similar to other models from the 'derestricted' family.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/GLM-4.7-Flash-Derestricted-GGUF --local-dir ./glm-4.7-flash-derestricted
llama-cli -m ./glm-4.7-flash-derestricted/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "koute/GLM-4.7-Flash-Derestricted"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve koute/GLM-4.7-Flash-Derestricted --trust-remote-codeDownloads and variants (4)
Every published repository of GLM-4.7-Flash-Derestricted, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/GLM-4.7-Flash-Derestricted-GGUF | GGUF | 130.8K |
| mradermacher/GLM-4.7-Flash-Derestricted-i1-GGUF | GGUF (imatrix) | 1.3K |
| koute/GLM-4.7-Flash-Derestrictedsource | Transformers | 156 |
| mlx-community/GLM-4.7-Flash-Derestricted-8bit-gs32 | MLX | 50 |
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-Derestricted 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 mit, but the terms that bind you are koute/GLM-4.7-Flash-Derestricted’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-Derestricted is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with koute, and does not independently verify publisher claims. See responsible use.