Qwen3.8-27B-Uncensored-MLX Demo

💬 Live Chat Demo

The demo below runs the PyTorch equivalent on ZeroGPU. For native MLX inference on your Mac, see the local usage section.

📊 Available Quantizations

8-bit~27.5 GB | 32 GB RAM
6-bit~22 GB | 24-32 GB RAM
4-bit~15 GB | 24 GB RAM
2-bit~8.7 GB | 16 GB RAM

✨ Key Features

ArchitectureHybrid Attention
Parameters27 Billion
Context262K tokens
VisionNative VL Tower
Tool CallingSupported

🎯 Quality vs BF16

8-bitcos 0.9997
6-bitcos 0.9996
4-bitcos 0.996
2-bitcos 0.92 ⚠️

🚀 Local Usage (MLX on Apple Silicon)

# Install mlx-vlm (requires mlx-vlm >= 0.6.13, mlx >= 0.32)
pip install -U mlx-vlm

# Download the 4-bit quantization (recommended)
hf download orcarouter/Qwen3.8-27B-Uncensored-MLX --include "4-bit/*" \
    --local-dir ./Qwen3.8-27B-Uncensored-MLX
                

# Text generation
python -m mlx_vlm generate \
    --model ./Qwen3.8-27B-Uncensored-MLX/4-bit \
    --prompt "Explain quantum entanglement in one sentence." \
    --max-tokens 256
                

# Vision (image + text)
python -m mlx_vlm generate \
    --model ./Qwen3.8-27B-Uncensored-MLX/4-bit \
    --image path/to/image.png \
    --prompt "Describe this image." \
    --max-tokens 256
                

# OpenAI-compatible server
python -m mlx_vlm server \
    --model ./Qwen3.8-27B-Uncensored-MLX/4-bit \
    --port 8080

# Then use with any OpenAI-compatible client
curl http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role": "user", "content": "Hello!"}]}'
                

⚠️ Disclaimer

This model has had its safety alignment substantially removed via abliteration. It is released strictly for legitimate research — interpretability, AI-safety and refusal-mechanism study, red-teaming, robustness evaluation, and controlled experiments. You assume full responsibility and liability for how you use it and for everything it generates.