GPT-oss-20b-HERETIC-uncensored
20BGPT-OSSheretic
from p-e-w/gpt-oss-20b-heretic
↓ 45.9K♥ 2027 repos
gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi
gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi is a Heretic-ablated variant of DavidAU/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4, published on Hugging Face by nightmedia. It is 20B parameters, 24GB VRAM class class and apache-2.0 licence. It is distributed across 2 repositories in MLX format, totalling 5.4K downloads.
Specification
- Model name
- gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi
- Publisher
- nightmedia
- Parameters
- 20B
- VRAM required
- ~13 GB at Q4_K_M, 4k context (estimated) — runs on 16 GB cards
- Hardware class
- 10B – 25B — 24GB VRAM class
- Licence
- apache-2.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- MLX
- First indexed
- 6 Feb 2026
- Last updated
- 6 Feb 2026
Ablation technique
How gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi was modified, and what that implies.
Heretic
Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.
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.
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "nightmedia/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4-q8-hi-mlx"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve nightmedia/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4-q8-hi-mlx --trust-remote-codeDownloads and variants (1)
The only published repository of gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi.
| Repository | Format | Downloads |
|---|---|---|
| nightmedia/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4-q8-hi-mlxsource | MLX | 5.4K |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi 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 DavidAU/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4, not added by ablation — at 20B 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.
- Declared licence is apache-2.0, but the terms that bind you are DavidAU/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4’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 gpt-oss-20B-INSTRUCT-Heretic-Uncensored-hi is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with nightmedia, and does not independently verify publisher claims. See responsible use.