🚀(Windows & Mac) Qwen3: Next-Generation Hybrid Large Language Model – Surpassing DeepSeek with Multi-Scale, Multi-Modal, and Multi-Language Support!

The Alibaba Cloud Qwen team leads innovation again with the release of Qwen3, their next-generation hybrid large language model! As a major upgrade from QwQ and Qwen2.5, Qwen3 offers model sizes ranging from 0.6B to 235B parameters, supporting both dense and MoE (Mixture of Experts) architectures to provide optimal solutions for various scenarios.

The flagship Qwen3 demonstrates exceptional performance in mathematics and coding metrics.

Comprehensively surpassing OpenAI o1 and o3 mini, DeepSeek R1, and Grok3.

🌟 Core Advantages of Qwen3

1. Flexible Model Size Options

  • 0.6B: Lightweight, suitable for mobile devices
  • 1.7B: Balanced performance and resource consumption
  • 4B: Suitable for general applications
  • 8B: Enhanced reasoning capabilities
  • 14B: Professional applications
  • 32B: Enterprise applications
  • 30B-A3B: Specially optimized version
  • 235B-A22B: Ultra-large scale version

2. Innovative Hybrid Thinking Modes

  • Thinking Mode: For complex problems, reaching accurate answers through step-by-step reasoning
  • Non-thinking Mode: Quick responses for simple questions, improving efficiency

3. Powerful Multi-language Support

  • Support for 119 languages and dialects
  • Global application scenario support
  • Seamless cross-language communication

4. Outstanding Reasoning Capabilities

  • Significantly improved mathematical reasoning
  • Enhanced code generation
  • More accurate common sense reasoning
  • More natural human-computer interaction

📥 Local Installation Guide

Windows Installation

1. Download Required Files

Download link: https://localai.top/95

Windows installation files include:

Filename Size VRAM Usage Purpose
0-OllamaSetup.exe 763,249 KB – Ollama main program installer for local deployment
1-Run-Qwen3-0.6B.bat 1 KB 523MB Startup script for Qwen3 0.6B model
1-Run-Qwen3-1.7B.bat 1 KB 1.4GB Startup script for Qwen3 1.7B model
1-Run-Qwen3-4B.bat 1 KB 2.6GB Startup script for Qwen3 4B model
1-Run-Qwen3-8B.bat 1 KB 5.2GB Startup script for Qwen3 8B model
1-Run-Qwen3-14B.bat 1 KB 9.3GB Startup script for Qwen3 14B model
1-Run-Qwen3-32B.bat 1 KB 20GB Startup script for Qwen3 32B model
2-Cherry-Studio-1.2.9-x64.exe – – Cherry Studio GUI (Windows version, optional)

2. Install Ollama (Windows)

  1. Double-click 0-OllamaSetup.exe
  2. Follow the installation wizard (click “Continue” as needed)
  3. Double-click the appropriate .bat script to start Ollama service and run the model

macOS Installation

1. Download Required Files

Mac installation files include:

Filename Purpose
0-Ollama-darwin.zip Ollama main program (Mac version)
1-Run-Qwen3-0.6B.command Startup script for Qwen3 0.6B model
1-Run-Qwen3-1.7B.command Startup script for Qwen3 1.7B model
1-Run-Qwen3-4B.command Startup script for Qwen3 4B model
1-Run-Qwen3-8B.command Startup script for Qwen3 8B model
1-Run-Qwen3-14B.command Startup script for Qwen3 14B model
1-Run-Qwen3-32B.command Startup script for Qwen3 32B model
2-Cherry-Studio-1.2.9-arm64.dmg Cherry Studio GUI (Mac version, arm64, optional)

2. Install Ollama (macOS)

  1. Extract 0-Ollama-darwin.zip
  2. Open Terminal and navigate to the extracted folder
  3. Run chmod +x 1-Run-Qwen3-*.command to grant execution permissions
  4. Double-click or run the appropriate .command script in Terminal

3. Choose and Run Model Scripts

Select the appropriate model based on your hardware configuration:

  • Entry-level hardware:

    • Recommended: 0.6B or 1.7B version
    • Minimum 8GB RAM
    • VRAM usage: 523MB – 1.4GB

  • Mid-range hardware:

    • Recommended: 4B or 8B version
    • Minimum 16GB RAM
    • VRAM usage: 2.6GB – 5.2GB

  • High-end hardware:

    • Recommended: 14B or 32B version
    • Minimum 32GB RAM
    • VRAM usage: 9.3GB – 20GB

4. Install Cherry Studio (Optional)

For a graphical interface:

  • Windows: Install 2-Cherry-Studio-1.2.9-x64.exe
  • Mac: Install 2-Cherry-Studio-1.2.9-arm64.dmg (supports Apple Silicon M1/M2/M3)

💡 Usage Recommendations

  1. Hardware-based Model Selection:

    • 8GB RAM: Use 0.6B or 1.7B
    • 16GB RAM: Use 4B or 8B
    • 32GB RAM: Use 14B
    • 64GB+ RAM: Use 32B

  2. Task-based Mode Selection:

    • Simple Q&A: Use non-thinking mode
    • Complex reasoning: Use thinking mode

  3. Multi-language Applications:

    • Support for 119 languages
    • Ideal for international business scenarios

🎯 Practical Application Scenarios

  1. Enterprise Applications:

    • Multi-language customer service systems
    • Intelligent document processing
    • Data analysis and report generation

  2. Personal Use:

    • Smart assistant
    • Learning aid
    • Creative writing
    • Code development

  3. Educational Field:

    • Multi-language teaching
    • Personalized tutoring
    • Knowledge Q&A

🔮 Future Outlook

The launch of Qwen3 marks a new milestone in large language model technology. By offering various sizes and modes, AI technology becomes more accessible and practical for real-world applications. Whether for individual users or enterprises, everyone can find the most suitable solution for their needs.

As technology continues to evolve, we look forward to seeing Qwen3 realize its potential in more fields, bringing greater convenience and innovation to human society.