A Python tool that analyzes your system's hardware and determines which AI models you can run locally.

• System Hardware Detection: Automatically detects CPU, RAM, GPU, and disk space
• AI Model Database: Fetches data on 200+ AI models from Ollama and other sources
• Compatibility Analysis: Determines which models your system can run based on resource requirements
• Detailed Reporting: Generates a comprehensive report with recommendations
• Support for Multiple GPUs: Detects and accounts for NVIDIA and AMD GPUs

• Python 3.7+
• Windows/Linux/macOS

psutil>=5.9.0
requests>=2.31.0
beautifulsoup4>=4.12.0
GPUtil>=1.4.0
wmi>=1.5.1 (Windows only)

1. Clone the repository:

git clone https://github.com/Ssenseii/ariana.git
cd ariana

2. Create a virtual environment (optional but recommended):

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Run the analyzer:

python main.py

The tool will:

1. Detect your system resources
2. Fetch available AI models
3. Analyze compatibility
4. Generate a detailed report (ai_capability_report.txt)

================================================================================
AI MODEL CAPABILITY ANALYZER
================================================================================

[1/4] Detecting system resources...
  ✓ CPU: 12 cores
  ✓ RAM: 31.11 GB available
  ✓ GPU: NVIDIA GeForce RTX 5060 Ti (15.93 GB VRAM)

[2/4] Fetching AI model data...
  ✓ Retrieved 217 AI models

[3/4] Analyzing model compatibility...
  ✓ Analysis complete
  ✓ You can run 158 out of 217 models

[4/4] Generating report...
  ✓ Report generated: ai_capability_report.txt

The analyzer detects your system specifications:

• CPU: Number of cores and max frequency
• RAM: Total and available memory
• GPU: VRAM and driver information
• Disk: Available space

Each AI model has specific requirements:

• Parameters: Model size (in billions)
• Quantization: Compression level (Q4, Q5, F8, etc.)
• RAM: Minimum RAM needed
• VRAM: Recommended GPU VRAM
• Disk: Space required to download

Models are scored based on:

• Available RAM vs. required RAM
• Available VRAM vs. recommended VRAM
• Available disk space
• Overall system capabilities

ai-model-capability-analyzer/
├── main.py                    # Entry point
├── system_analyzer.py         # Hardware detection
├── model_scraper.py          # Fetches model data
├── analyzer.py               # Compatibility analysis
├── report_generator.py       # Report generation
├── requirements.txt          # Project dependencies
└── README.md                 # This file

The analyzer supports 200+ models including:

• Base Models: Llama, Mistral, Qwen, Gemma, Phi, Falcon
• Specialized Models: CodeLlama, DeepSeek-Coder, StarCoder
• Vision Models: LLaVA, Qwen-VL, MiniCPM-V
• Embedding Models: nomic-embed-text, mxbai-embed-large

After analyzing your system, the tool provides recommendations:

• Which models to try first
• Tips for optimizing performance
• GPU acceleration options
• Best practices for running models locally

1. Start Small: Begin with smaller models (0.5B-1B parameters)
2. Use Quantization: Q4/Q5 quantization significantly reduces memory usage
3. GPU Acceleration: Enable GPU acceleration for faster inference
4. Use Ollama: Simplifies model management and deployment
5. Monitor Resources: Watch CPU/RAM usage while running models

• VRAM detection may not be accurate on all systems
• Windows with WMI may have detection issues in some configurations
• GPU detection is optimized for NVIDIA and AMD GPUs
• Model requirements are estimates based on typical scenarios

• Ensure your GPU drivers are up to date
• Install GPU detection libraries for your hardware

• Use smaller models
• Apply higher quantization (Q4 instead of F16)
• Close other applications

• Results are estimates based on available specifications
• Actual performance depends on model optimization and driver versions

This project is licensed under the MIT License - see the LICENSE file for details.

Contributions are welcome! Please feel free to submit a Pull Request.

• Model data sourced from Ollama
• Inspired by the need to understand AI model compatibility with consumer hardware

This tool provides estimates based on theoretical requirements. Actual performance may vary depending on:

• System optimization
• Driver versions
• Background processes
• Model-specific optimizations

Always test models on your system before relying on them for production use.