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

Bigbarry/abliterated-agent

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/Bigbarry%2Fabliterated-agent"
Response includes
  • classification m1
  • files 3
  • author_summary 1 models
  • readme_text full
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)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
0
Likes
0
Model age
3mo ago
created 2026-07-04
Downloads over time
Now0→from0↑0%
00110 on Jul 150 on Oct 12JulAugSepOct
Jul 15 → Oct 12 · 54 snapshots · spans 89 days

Metadata

Tags
region:us
Total size
0 B
Files
3
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-07-04 15:54

Files by quantization

Auxiliary files 3 files 22.2 KB
abliterated_agent.py 17.9 KB 0d1e8346 download
README.md 2.80 KB 4e7b537e download
.gitattributes 1.48 KB a6344aac download

README current version from Hugging Face

Abliterated Tool-Use Agent

A Python script that runs an uncensored, abliterated Qwen3 model via Ollama with full tool-use capabilities.

What it does

The model can autonomously:

  • Search the web (DuckDuckGo, no API key needed)
  • Fetch and read webpages (extract text from any URL)
  • Run shell commands on your machine (nmap, curl, scripts, whatever)
  • Read and write files on disk
  • Make HTTP requests to any API or endpoint
  • Then synthesize everything into a concise answer — no rambling

Quick start — Kali PC

# 1. Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# 2. Download the 8B model (smarter, 4.7GB)
ollama pull huihui/qwen3-abliterated:8b

# 3. Install Python deps
pip install requests beautifulsoup4

# 4. Download this script
wget -O abliterated_agent.py https://huggingface.co/Bigbarry/abliterated-agent/resolve/main/abliterated_agent.py

# 5. Run it
python3 abliterated_agent.py

Quick start — Phone (NetHunter / Termux)

# 1. Install Ollama
pkg install ollama

# 2. Start the server
ollama serve &

# 3. Download the 4B model (smaller, 2.3GB, fits on phone)
ollama pull huihui/qwen3-abliterated:4b-instruct-2507

# 4. Install Python deps
pip install requests beautifulsoup4

# 5. Download this script
wget -O abliterated_agent.py https://huggingface.co/Bigbarry/abliterated-agent/resolve/main/abliterated_agent.py

# 6. Run it with the 4B model
python3 abliterated_agent.py --model huihui/qwen3-abliterated:4b-instruct-2507

Usage

# Interactive chat (default)
python3 abliterated_agent.py

# Use the smaller 4B model (better for phone)
python3 abliterated_agent.py --model huihui/qwen3-abliterated:4b-instruct-2507

# Disable shell execution (safer mode)
python3 abliterated_agent.py --no-shell

# Run a single prompt (no interactive mode)
python3 abliterated_agent.py --prompt "search the web for latest CVEs and summarize"

Models

Model Size Best for Ollama tag
Qwen3-4B-Instruct-2507 abliterated 2.3GB (Q4_K_M) Phone / NetHunter huihui/qwen3-abliterated:4b-instruct-2507
Qwen3-8B abliterated v2 4.7GB (Q4_K_M) Kali PC / laptop huihui/qwen3-abliterated:8b

The 4B "Instruct-2507" variant is non-thinking — it gives direct, concise answers without chain-of-thought rambling.

Tools available

Tool What it does
web_search Search DuckDuckGo, returns titles/URLs/snippets
fetch_webpage Download a URL and extract readable text
http_request Make GET/POST/PUT/DELETE requests to any URL
run_shell Execute shell commands, returns stdout/stderr/exit code
read_file Read any file from disk
write_file Write content to a file on disk

License

Apache 2.0 (model) / script is free to use.

README history 1 version

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

  1. 2026-07-04Upload README.mdeccea472.8 KB
    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.

Open in Abliteration