Gpt Oss 20b RichardErkhov Heresy needs about 13.7 GB of memory at IQ4_XS (a 12.8 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 3 quants, from 12.7 GB (MXFP4_MOE) to 13.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 1.6 GB.
20.9B‑parameter gpt‑oss fine‑tune with high willingness score
The gpt‑oss‑20b‑RichardErkhov‑heresy model is a 20.9B‑parameter fine‑tune of the gpt‑oss family. It was produced through a scuffed modification of P‑E‑W’s Heretic abliteration engine with Magnitude‑Preserving Orthogonal Ablation. The model is licensed under Apache‑2.0 and supports a 131072‑token context.
Good at
- High‑reasoning tasks
- Agentic use cases
- Versatile developer scenarios
Limits
- No vision support
- Requires harmony response format
- Large VRAM requirement for full precision
Best for
- Local inference on high‑VRAM GPUs
- Custom agent development
- Research on open‑weight models
Who it’s for. Developers and researchers who need a powerful, open‑weight GPT‑style model for reasoning and agentic tasks. The Apache‑2.0 license allows free experimentation and commercial deployment.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| MXFP4_MOE | 12.7 GB | 13.6 GB | Too big | Get file → | |
| IQ4_XS | 12.8 GB | 13.7 GB | Too big | Get file → | |
| BF16 | 13.8 GB | 14.7 GB | Too big | Get file → |
For 8 GB VRAM, use the IQ4_XS quant (12.79 GB file, ~13.7 GB memory at 8 k context). For 12 GB VRAM, MXFP4_MOE (12.65 GB file, ~13.6 GB memory). For 16 GB VRAM, BF16 (13.79 GB file, ~14.7 GB memory). Estimates are approximate.
What GPU do you need for Gpt Oss 20b RichardErkhov Heresy?
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: the best quant that fits is BF16 (13.8 GB file, about 14.7 GB in use).
Which Gpt Oss 20b RichardErkhov Heresy quant should you choose?
IQ4_XS (12.8 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is MXFP4_MOE at 12.7 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about Gpt Oss 20b RichardErkhov Heresy
How much VRAM does Gpt Oss 20b RichardErkhov Heresy need?
About 13.7 GB at IQ4_XS with an 8K context window. That is the 12.8 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 RichardErkhov Heresy on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 13.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Gpt Oss 20b RichardErkhov Heresy GGUF?
It depends on the quant: MXFP4_MOE is 12.7 GB, IQ4_XS is 12.8 GB, BF16 is 13.8 GB.
What context length does Gpt Oss 20b RichardErkhov Heresy 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 RichardErkhov Heresy GGUF?
The files are on Hugging Face in the MuXodious/gpt-oss-20b-RichardErkhov-heresy-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 RichardErkhov Heresy 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?
Yes, using the IQ4_XS quant which requires about 13.7 GB memory at 8 k context.
Does the model support image inputs?
No, the model has no vision capabilities.
What license does it use?
Apache‑2.0, allowing free use and modification.