# Pungyo Twin — Run Guide (stability profile)

Run a Qwen3.6 35B-A3B class GGUF on a single 8GB GPU with a stability-first
profile. You provide the model. Pungyo is a single self-contained binary.

## 1. Get a model

Download a compatible Qwen3.6 35B-A3B GGUF from the official source:
https://huggingface.co/Qwen

Pungyo does not ship, modify, or relicense model weights.

## 2. Unzip

```
Pungyo-Qwen36-Twin-Public-Verification-20260530.zip
├── bin\pungyo.exe
├── bin\cublas64_13.dll
└── bin\cublasLt64_13.dll
```

## 3. Run (twin / stability mode)

```powershell
.\bin\pungyo.exe "D:\models\Qwen3.6-35B-A3B.gguf" "Hello, my name is" wireframe
```

The `wireframe` argument selects the twin (stability) profile. No setup, no
separate llama.cpp install. Operating point chosen in memory, never on the
command line.

## Single vs Twin

- omit the 3rd argument (or pass `solid`) → maximum speed, tight VRAM.
- pass `wireframe` (3rd argument) → stability-first, more VRAM headroom, fewer
  OOM under load.

## Notes

- Requires Windows, NVIDIA RTX-class 8GB GPU with a recent NVIDIA driver.
- 32 GB system RAM recommended.
- Speed is a local observation and varies by hardware.
- The mechanism is not documented here by design (patent-filed).
- Issues / requests: use the repository Community tab.