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

Shiftedx/ornith-1.0-9b-abliterated-mxfp4-mlx

Shiftedx 9.0B
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/Shiftedx%2Fornith-1.0-9b-abliterated-mxfp4-mlx"
Response includes
  • classification m1
  • files 10
  • hub_downloads_all_time 1,133
  • author_summary 24 models
  • readme_text full
Try next
GET/models?author=Shiftedx
All 24 models by Shiftedx
 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
1K
51 last 30d - cooling
Likes
0
Model age
3mo ago
created 2026-07-05
Downloads over time
Now1.1K→from737↑55%
7178711K1.2K737 on Jul 151.1K on Oct 111.1K on Oct 9JulAugSepOct
Jul 15 → Oct 11 · 53 snapshots · spans 88 days

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.

deepreinforce-ai/Ornith-1.0-9B this lineage
Shiftedx/ornith-1.0-9b-abliterated-mxfp4-mlx↓ 51

Metadata

License
other
Languages
en
Tags
mlx safetensors qwen3_5 apple-silicon ornith mxfp4 abliterated en base_model:ornith-ai/Ornith-1.0-9B base_model:quantized:ornith-ai/Ornith-1.0-9B license:other 4-bit

Related

Other by Shiftedx
qwen3.8-27b-aeon-ultimate-uncensored-attention8-bf16recurrence-vision-mtplx 1K qwen3.8-27b-aeon-ultimate-uncensored-mxfp4-vision-mtplx 1K Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-MLX-MIXED-VISION-MTPLX 543 qwopus3.8-27b-flash-abliterated-mxfp4-vision-mtplx 541 ornith-1.5-35b-a3b-abliterated-attention8-bf16recurrence-vision-mtplx 474
Same base model
huihui-ai/Huihui-Ornith-1.0-9B-abliterated-MTP-GGUF 6K zaakirio/Ornith-1.0-9B-uncensored-GGUF 650 maci0/Ornith-1.0-9B-abliterated-NVFP4 386 Dzluck/Ornith-1.0-9B-abliterated-MTP-GGUF 140 PeppX/Ornith-1.0-9B-Uncensored 60
Total size
4.43 GB
Files
10
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-07-05 23:07
Source huggingface.co/Shiftedx/ornith-1.0-9b-abliterated-mxfp4-mlx ↗ Base model deepreinforce-ai/Ornith-1.0-9B ↗

Files by quantization

Auxiliary files 10 files 4.46 GB
model.safetensors 4.43 GB 10c32dcb download
tokenizer.json 19.1 MB 06b95093 download
vocab.json 6.41 MB 0aa0ce06 download
model.safetensors.index.json 57.9 KB 8efa825e download
tokenizer_config.json 8.84 KB 2b9dc3a7 download
chat_template.jinja 7.42 KB 84be41b3 download
config.json 3.23 KB 2d5e8224 download
.gitattributes 1.53 KB 52373fe2 download
README.md 973 B 5e9db6aa download
generation_config.json 176 B 0534edd0 download

README current version from Hugging Face

language:

  • en
    license: other
    tags:
  • mlx
  • apple-silicon
  • ornith
  • mxfp4
  • abliterated
    base_model: deepreinforce-ai/Ornith-1.0-9B

ornith-1.0-9b-abliterated-mxfp4-mlx

MXFP4 MLX build of deepreinforce-ai/Ornith-1.0-9B with a local abliteration pass to reduce over-refusal while preserving normal task behavior.

Notes

  • Publisher: Shiftedx
  • Format: MLX safetensors
  • Quantization: MXFP4
  • Vision: no
  • Default chat behavior: thinking off unless explicitly enabled
  • Privacy: local-first artifact; no service endpoint, telemetry, private prompts, or host-specific paths are included

Validation

Hardware reference: Apple M4 Max Apple Silicon with 64 GB unified memory.

  • Refusal-marker mini-suite: 0% after abliteration; baseline was 100%.
  • Benign-sensitive refusal rate: 0%.
  • Utility mini-suite: 100%.
  • Code mini-suite: 100%.
  • LM Studio load and no-loop text smoke: passed.

Use

Load with LM Studio or MLX-compatible tooling on Apple Silicon.

README history 2 versions

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

  1. 2026-07-05Add validation hardware reference8a4b90e973 B
    Loading...
  2. 2026-07-05Upload validated abliterated MLX MXFP4 buildd02b463898 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