GP

gpt oss 20b abliterated

gpt-ossGGUF by wangzhangUpdated 2026-08-29
Download on Hugging Face
Parameters21B
Context128K
Layers24
Architecturegpt-oss
Smallest quant15.8 GB
Quick answer

gpt oss 20b abliterated needs about 16.7 GB of memory at Q4_K_M (a 15.8 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 3 quants, from 15.8 GB (Q4_K_M) to 41.9 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 1.6 GB.

gpt oss 20b abliterated – a 21‑billion‑parameter MoE model with refusal suppression

The gpt oss 20b abliterated model is a 21‑billion‑parameter GPT‑OSS variant that has undergone direct weight editing and MoE router suppression to reduce refusal rates. It is released under the apache‑2.0 license and is intended for safety research and mechanism analysis.

Good at

  • Reduced refusal on harmful prompts
  • Direct weight‑edited MoE architecture
  • Compatible with GGUF runtimes

Limits

  • Requires large VRAM – Q4_K_M needs ~16.7 GB at 8 k context
  • No vision or audio support
  • No official benchmark scores provided

Best for

  • Red‑team testing of safety mitigations
  • Mechanism analysis of MoE routers
  • Research on refusal suppression techniques

Who it’s for. Researchers and developers working on AI safety, red‑team evaluation, and MoE mechanism studies. The model is not intended for general content generation or distribution of harmful content.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q4_K_M15.8 GB 16.7 GB
Too big Get file →
Q8_022.3 GB 23.2 GB
Too big Get file →
BF1641.9 GB 42.8 GB
Too big Get file →

For 16 GB VRAM, use the Q4_K_M quantised file (15.81 GB, ~16.7 GB memory). If you have 24 GB VRAM, Q8_0 (22.26 GB, ~23.2 GB memory) is available. BF16 requires 42 GB memory and is only for high‑end setups.

What GPU do you need for gpt oss 20b abliterated?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: the best quant that fits is Q8_0 (22.3 GB file, about 23.2 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (22.3 GB file, about 23.2 GB in use).
  • 48 GB VRAM: the best quant that fits is BF16 (41.9 GB file, about 42.8 GB in use).

Which gpt oss 20b abliterated quant should you choose?

Q4_K_M is the usual starting point for gpt oss 20b abliterated: it balances file size and quality well. Here it is a 15.8 GB file.

The Q8_0 file is 22.3 GB, about 1.4 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegpt-oss
Parameters21B (approximate, mixture of experts)
Layers24
Trained context131,072 tokens
KV cache (fp16)48 KB per token

Source

Quantized bywangzhang
Familygpt-oss
Licenseapache-2.0
Downloads (HF)1,871
Last updated2026-08-29

Frequently asked questions about gpt oss 20b abliterated

How much VRAM does gpt oss 20b abliterated need?

About 16.7 GB at Q4_K_M with an 8K context window. That is the 15.8 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run gpt oss 20b abliterated on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 16.7 GB, so you would need a larger GPU or CPU offloading.

What is the file size of gpt oss 20b abliterated GGUF?

It depends on the quant: Q4_K_M is 15.8 GB, Q8_0 is 22.3 GB, BF16 is 41.9 GB.

What context length does gpt oss 20b abliterated support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.4 GB at 8K and 1.6 GB at 32K.

Where can I download gpt oss 20b abliterated GGUF?

The files are on Hugging Face in the wangzhang/gpt-oss-20b-abliterated-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does gpt oss 20b abliterated use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run this model on 8 GB VRAM?

No. Even the smallest quantised variant requires at least 16.7 GB of VRAM at 8 k context.

What license applies to this model?

The model is released under the apache‑2.0 license, inherited from the upstream OpenAI gpt‑oss release.

What does "abliterated" mean in this context?

It refers to direct weight editing and MoE router suppression that reduce refusal rates on harmful prompts.

Other GGUFs of gpt-oss-20b-abliterated

Similar models

Browse all →

More from wangzhang

All by wangzhang →

More gpt-oss models

All gpt-oss →

Other medium models

All medium models →