Get the app
← back to catalog · registered 2026-08-22 13:56

batsclamp/Huihui-gpt-oss-120b-mxfp4-abliterated

batsclamp Gpt-oss 118B second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model. 

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/batsclamp%2FHuihui-gpt-oss-120b-mxfp4-abliterated"
Response includes
  • classification m1
  • files 24
  • benchmarks 11 entries
  • hub_downloads_all_time 7,329
  • author_summary 7 models
  • readme_text full
Try next
GET/models?author=batsclamp
All 7 models by batsclamp
 GET/models?family=gpt-oss&method=m1
Sibling gpt-oss models · same method
 GET/methods/m1
What "m1" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
How we label → About M1 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
7K
362 last 30d - cooling
Likes
5
Model age
7mo ago
created 2026-02-17
Downloads over time
Now7.4K→from51↑14,478%
02.7K5.4K8.2K51 on Feb 187.4K on Oct 11FebAprJunAugOct
Feb 18 → Oct 11 · 73 snapshots · spans 235 days

Benchmarks

Benchmark Score Source
Entertainment 1.4 UGI
Hazardous 1.2 UGI
Natural Intelligence 16.72 UGI
Political lean -1.2% UGI
Sensitive-Info 12.89 UGI
SocPol 1.3 UGI
UGI 26.09 UGI
Willingness (10) 5.2 UGI
W10-Adherence 3.5 UGI
W10-Direct 7 UGI
Writing NA UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

huihui-ai/Huihui-gpt-oss-120b-BF16-abliterated this lineage
batsclamp/Huihui-gpt-oss-120b-mxfp4-abliterated↓ 362

Metadata

Tags
safetensors gpt_oss 8-bit mxfp4 region:us

Related

Other by batsclamp
Huihui-Qwen3.8-27B-abliterated-FP8-v2 2K Huihui-Qwen3.6-35B-A3B-abliterated-FP8 595 Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-FP8 394 Huihui-Qwen3.6-35B-A3B-Claude-4.6-Opus-abliterated-FP8 142 Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated-FP8 102
Same base model
mradermacher/Huihui-gpt-oss-120b-BF16-abliterated-i1-GGUF 56 mradermacher/Huihui-gpt-oss-120b-BF16-abliterated-GGUF 0
Total size
60.8 GB
Files
24
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-03-27 13:12
Source huggingface.co/batsclamp/Huihui-gpt-oss-120b-mxfp4-abliterated ↗ Base model huihui-ai/Huihui-gpt-oss-120b-BF16-abliterated ↗

Files by quantization

Auxiliary files 24 files 60.8 GB
model-00001-of-00015.safetensors 4.38 GB d972df5d download
model-00006-of-00015.safetensors 4.31 GB 7648d923 download
model-00008-of-00015.safetensors 4.31 GB ff9d2b11 download
model-00010-of-00015.safetensors 4.31 GB 6019b0da download
model-00012-of-00015.safetensors 4.31 GB 88647094 download
model-00014-of-00015.safetensors 4.31 GB 1a4c517e download
model-00002-of-00015.safetensors 4.31 GB c0f33017 download
model-00004-of-00015.safetensors 4.31 GB e5e21c16 download
model-00005-of-00015.safetensors 3.83 GB 517d85dc download
model-00007-of-00015.safetensors 3.83 GB ff77230d download
model-00009-of-00015.safetensors 3.83 GB 444f2e10 download
model-00011-of-00015.safetensors 3.83 GB 3871dd0f download
model-00013-of-00015.safetensors 3.83 GB 507a8640 download
model-00003-of-00015.safetensors 3.83 GB c13d6a71 download
model-00015-of-00015.safetensors 3.23 GB 10f811d2 download
tokenizer.json 26.6 MB 0614fe83 download
model.safetensors.index.json 56.0 KB 54eb081c download
chat_template.jinja 16.3 KB dc7bb119 download
tokenizer_config.json 4.10 KB c021cddb download
config.json 2.33 KB 9ec50b8a download
.gitattributes 1.53 KB 52373fe2 download
README.md 844 B 7077f7d8 download
special_tokens_map.json 440 B 6274cc1b download
generation_config.json 172 B b27df957 download

README current version from Hugging Face

base_model:

  • huihui-ai/Huihui-gpt-oss-120b-BF16-abliterated

Original model - https://huggingface.co/huihui-ai/Huihui-gpt-oss-120b-BF16-abliterated

This model was specially built to be used with Nvidia DGX Spark

Abliterated version of OSS-GPT-120B converted back from BF16 to MXFP4 (original format) so that you can use it natively with vLLM.

Gives you 40-50 tps with 100k context on Nvidia Spark DGX, which makes it a perfect choice for a the daily driver.

Please be aware - it's "jailbroken" / "abliterated" / "unlocked" model! it will do whatever you tell it to do, so don't use it unless you know what you are doing!

there are alternative models too:

  • justinjja/gpt-oss-120b-Derestricted-MXFP4
  • kldzj/gpt-oss-120b-heretic-v2

both are not as "derestricted" as this one, but it is possible that they are more "accurate".

README history 5 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-03-27Update README.md4be0d67844 B
    Loading...
  2. 2026-02-23Update README.mdc403b34775 B
    Loading...
  3. 2026-02-23Update README.mdc465017568 B
    Loading...
  4. 2026-02-17Update README.md74eb7d2567 B
    Loading...
  5. 2026-02-17Create README.md87d3cf6335 B
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration