Qwen3 Setup - Claude Code Skill

Get Qwen3 running locally with one command. This skill detects your system, recommends the optimal model size, and handles the entire installation.

Qwen3 is a state-of-the-art open model that runs well on consumer hardware:

• Efficient: Multiple sizes from 0.6B to 32B parameters
• Capable: Excellent reasoning, coding, and instruction following
• MoE option: 30B-A3B uses mixture-of-experts for efficiency
• Well-supported: First-class ollama support

Install the qwen3 skill from https://github.com/getaifs/qwen3-skill

git clone https://github.com/getaifs/qwen3-skill.git ~/.claude/skills/qwen3

# 1. Check your system
python3 ~/.claude/skills/qwen3/setup.py status

# 2. Install (auto-detects best model for your hardware)
python3 ~/.claude/skills/qwen3/setup.py install

# 3. Use it!
ollama run qwen3:8b "Hello!"

1. Detects your platform: macOS, Linux, or WSL2
2. Checks your memory: RAM and VRAM (for NVIDIA GPUs)
3. Recommends a model: Based on what your system can handle
4. Installs ollama: Via Homebrew (macOS) or official installer (Linux)
5. Pulls the model: Downloads and sets up Qwen3
6. Verifies it works: Runs a test prompt

The script automatically recommends a model based on your available memory:

Override with --model:

python3 ~/.claude/skills/qwen3/setup.py install --model qwen3:8b

# Check system and current status
python3 ~/.claude/skills/qwen3/setup.py status

# Full installation (recommended)
python3 ~/.claude/skills/qwen3/setup.py install

# Install specific model
python3 ~/.claude/skills/qwen3/setup.py install --model qwen3:14b

# Pull model only (ollama must be running)
python3 ~/.claude/skills/qwen3/setup.py pull

# Verify Qwen3 is working
python3 ~/.claude/skills/qwen3/setup.py verify

==================================================
QWEN3 SETUP - System Status
==================================================

Platform: macos (Apple Silicon)
RAM: 48.0 GB

Recommended model: qwen3:30b-a3b
  Parameters: 30B-A3B (MoE)
  Memory needed: 18.0 GB

Model options:
  ✓ qwen3:0.6b      (0.6B      ) - 0.8 GB
  ✓ qwen3:1.7b      (1.7B      ) - 1.5 GB
  ✓ qwen3:4b        (4B        ) - 3.0 GB
  ✓ qwen3:8b        (8B        ) - 5.5 GB
  ✓ qwen3:14b       (14B       ) - 9.5 GB
  ✓ qwen3:32b       (32B       ) - 20.0 GB
  ✓ qwen3:30b-a3b   (30B-A3B (MoE)) - 18.0 GB

Ollama installed: Yes
Ollama running: Yes
Qwen models: qwen3:8b

--------------------------------------------------
Status: READY ✓
Qwen3 is installed and ready to use.

This skill provides the local LLM foundation for:

• facts-from-openai - Extract personal facts from ChatGPT exports
• local-llm - Privacy-preserving document processing

• Python 3.8+
• macOS, Linux, or WSL2
• Internet connection (for downloading ollama and model)
• Sufficient RAM/VRAM for your chosen model size

MIT License - See LICENSE for details.