4. System Requirements & Installation
System Requirements
Lite Edition (BYO Model)
Operating System:
- Windows 10 (64-bit) or Windows 11
- macOS 12 (Monterey) or later (Apple Silicon or Intel)
Hardware:
- CPU: Modern x86-64 or Apple Silicon
- RAM: 16GB minimum
- Disk Space: ~5GB for the base install plus 2-8GB per local model
- Display: 1280x720 or higher
With these specs, you can:
- Run small-to-medium GGUF models (1-4B parameters)
- Profile and query datasets that fit in available RAM
- Use the Stage 2 connector to bring cloud AI in for heavier reasoning
Full Edition (Bundled Local LLM)
Hardware:
- CPU: 8+ cores, modern x86-64, or Apple Silicon (M1+)
- RAM: 32GB+ on x86, or 16GB+ unified memory on Apple Silicon
- Disk Space: 50GB+ free (room for multiple models)
- Display: 1920x1080 or higher
Optional GPU (Highly Recommended on x86):
- NVIDIA: 8GB+ VRAM (RTX 3060 12GB or better recommended)
- AMD on Windows: 8GB+ VRAM via Vulkan
- Apple Silicon: Metal acceleration built in — no separate GPU needed
With these specs, you can:
- Run larger models (7-14B parameters) locally with reasonable latency
- Process larger datasets
- Reduce dependence on cloud AI for sensitive workloads
High-Performance Setup
For maximum performance:
- CPU: 16+ cores (Ryzen 9, Intel i9, Xeon)
- RAM: 32GB+ (64GB ideal for very large datasets)
- GPU: RTX 4090, A6000, or Apple M3 Max/Ultra
- Disk: NVMe SSD with 100GB+ free
Performance gains:
- Run multiple large models simultaneously
- Process 50GB+ datasets
- Sub-second query responses
- Batch processing of thousands of files
GPU Recommendations
NVIDIA GPUs
Entry Level (Budget):
- GTX 1060 6GB - Can run smaller models (Q4)
- RTX 3060 12GB - Good for medium models (Q5, Q8)
Mid-Range (Recommended):
- RTX 3080 10GB - Excellent performance
- RTX 4070 12GB - Great efficiency
- RTX 4080 16GB - High performance
High-End (Professional):
- RTX 4090 24GB - Maximum consumer performance
- A6000 48GB - Workstation/server grade
- H100 80GB - Data center (overkill for RangerIO)
AMD GPUs (Windows, via Vulkan)
Entry Level:
- RX 6600 XT 8GB — basic acceleration
- RX 6800 XT 16GB — solid mid-range performance
Mid-Range:
- RX 7800 XT 16GB — current generation
High-End:
- RX 7900 XTX 24GB — top consumer performance
Apple Silicon
Entry Level:
- M-series base chip with 16GB+ unified memory
Recommended:
- M-series Pro/Max with 16GB+ unified memory
- More unified memory = larger local models can fit
High-End:
- Ultra chip with 32GB+ unified memory
- Power-efficient inference for frequent local AI use
CPU Fallback
RangerIO ships a CPU path on every supported platform. If no compatible GPU is detected, the inference engine falls back to CPU automatically. This is slower but always available — useful on laptops without discrete graphics or on Windows machines without NVIDIA/AMD GPUs.
Disk Space Breakdown
Base Installation: ~5GB
- Application binaries
- System libraries
- Base configuration
AI Models: Variable
- Small models (1-3B): 1-2GB each
- Medium models (7B): 4-5GB each
- Large models (13B+): 8-15GB each
Example configurations:
- Minimal (1 model): ~7GB total
- Standard (3 models): ~15GB total
- Comprehensive (6+ models): ~30GB+ total
Data Storage: User-Dependent
- Project metadata: ~10MB per project
- Vector embeddings: ~1KB per document chunk
- Cached data: Varies by dataset size
Planning guide:
- Small projects (<1GB data): Add 2GB workspace
- Medium projects (1-10GB): Add 20GB workspace
- Large projects (10GB+): Add 3x your data size
Installation
Download
Visit the RangerIO website to download the installer for your platform:
- Windows: RangerIO-Setup.exe
- macOS: RangerIO.dmg
Installation Steps
Windows
- Run installer: Double-click
RangerIO-Setup.exe - Accept UAC prompt (requires admin rights)
- Choose install location (default:
C:\Program Files\RangerIO) - Wait for installation (~2 minutes)
- Launch RangerIO from Start Menu or desktop shortcut
First launch:
- Windows Defender may scan the application (normal)
- Firewall prompt: Allow local-only access (the backend runs on localhost)
macOS
- Open DMG: Double-click
RangerIO.dmg - Drag to Applications folder
- First launch: Right-click → Open (to bypass Gatekeeper if the build isn't signed for your machine)
- Accept security prompt
- RangerIO opens as a desktop application
Troubleshooting:
- If blocked: System Settings → Privacy & Security → "Open Anyway"
- Apple Silicon Macs: Native arm64 binary (faster than Intel build under Rosetta)
First Launch
Initial Setup Wizard
When you first launch RangerIO, you'll see a setup wizard guiding you through initial configuration:
💡 First Time Setup: The wizard appears automatically on first launch. You can access settings later from the gear icon ⚙️ in the top-right corner
Step 1: Welcome
- Brief introduction
- System requirements check
- License agreement (if applicable)
Step 2: Model Download (Full edition)
RangerIO classifies your hardware and suggests a bundled model that fits. Typical picks:
- Light hardware: Qwen3 1.7B or Phi-3.5-mini (~1-3GB)
- Standard hardware: Qwen3 4B (~2-3GB)
- High-RAM / GPU: Qwen3 8B or DeepSeek-R1 Distill 7B (~5GB)
- Skip: Use the Stage 2 connector instead, or drop a GGUF in later
The Lite edition does not bundle a model — you bring your own GGUF or use the Stage 2 connector.
Step 3: Data Location
Choose where to store:
- Projects and data
- Model files
- Temporary files
- Default: ~/RangerIO (or C:\Users\<You>\RangerIO)
Step 4: Optimization Profile
RangerIO detects your hardware and suggests a profile:
- CPU Only - No compatible GPU detected
- NVIDIA GPU - CUDA acceleration enabled
- AMD on Windows - Vulkan acceleration enabled
- Apple Silicon - Metal acceleration enabled
Step 5: Ready
- Configuration summary
- Launch application
Model Download Process
If you chose to download models:
1. Select models from the list
2. Download starts (progress bar shown)
3. Verification (checksum validation)
4. Installation (model extracted to storage location)
5. Ready (model available for use)
Download speeds:
- Local models: Instant (if pre-downloaded)
- Internet: Varies by connection (models are 1-8GB each)
Offline installation:
- Download models separately
- Place in ~/RangerIO/models/ directory
- RangerIO auto-detects them
Configuration
Settings Overview
Access settings from the gear icon in the top-right:
General:
- Default project location
- Language preference
- Theme (light/dark)
Models:
- Installed models list
- Download new models
- Set default model
- Configure context limits
Performance:
- Optimization profile
- GPU selection (if multiple)
- Memory limits
- CPU thread count
Data:
- Default import settings
- Data retention policies
- Cache management
Privacy:
- Telemetry (always off by default)
- Crash reporting (optional, local only)
- Logging level
Advanced Configuration
For power users, RangerIO stores configuration in:
- Windows: %LOCALAPPDATA%\RangerIO\
- macOS: ~/Library/Application Support/RangerIO/
Updating RangerIO
Check for Updates
RangerIO does NOT automatically check for updates (privacy by design).
Manual check:
1. Help → About RangerIO
2. Current version displayed
3. Visit website to check for newer version
Update Process
- Download new installer from website
- Close RangerIO
- Run new installer (overwrites old version)
- Data preserved (projects, models, settings)
- Launch updated version
Note: Your projects and models are NOT affected by updates.
Uninstalling
Windows
- Control Panel → Programs and Features
- Select RangerIO → Uninstall
- Choose to keep or remove data directory
macOS
- Drag RangerIO from Applications to Trash
- Manually remove:
~/Library/Application Support/RangerIO(optional)
Troubleshooting Installation
Common Issues
"Application can't be opened" (macOS):
- Right-click → Open (first time only)
- System Preferences → Security → "Open Anyway"
"Windows Protected your PC" (Windows):
- Click "More info"
- Click "Run anyway"
GPU not detected:
- Update graphics drivers
- For NVIDIA: confirm a recent CUDA-capable driver is installed
- For AMD on Windows: confirm a recent Vulkan-capable driver is installed
- Restart after driver update
Model download fails:
- Check internet connection
- Verify disk space
- Try downloading manually from website
Port 9000 / 5173 already in use:
- The backend runs on 9000 and (in dev) the frontend on 5173
- Another application may be using these ports — close it or change RangerIO ports in config
Getting Help
If you encounter issues:
1. Check the FAQ section
2. Review logs: Help → Open Logs Folder
3. Visit community forum
4. Contact support with log files