System Requirements & Device Compatibility

This document outlines the essential system requirements and device compatibility for our powerful, private AI platform. Meeting these requirements is crucial for ensuring optimal performance and a seamless experience with our products.

If you encounter any errors or unexpected behavior, please verify that your system and devices meets the specifications detailed in this guide. Adhering to these requirements will help prevent issues and ensure our product functions correctly on your devices.

Basic Requirements

Operating System

• macOS 15+ is required

Storage

• Available disk space for installation and AI models
• A SSD is recommended for better performance

Network

• Internet connection for initial setup and model downloads
• Ongoing internet is optional after setup (webAI runs locally)
• Wi-Fi (802.11ac/ax) and Ethernet (1Gb/10Gb) for local network discovery and communication.
• IPv4 Protocols are required.
• P2P traffic must be allowed on the network for nodes to communicate.

Hardware

• webAI products are currently exclusively supported on Apple devices featuring Apple Silicon M-series processors (e.g., M1, M2, M3, M4, M5).
• 256GB+ of Disk space recommended (larger AI models will require more disk space).

• 16GB+ of RAM recommended (larger AI models will require more memory).

  On macOS, applications can only access a portion of system memory for GPU operations. macOS limits GPU usage to approximately 60% of available RAM, which may affect performance for memory-intensive workloads.

Security Configuration

• Exceptions: To ensure proper functionality, endpoint security software must include exceptions for the following directories:

  .webai

  .cache

Memory

• Sufficient RAM for the AI models you plan to use

System Definitions

Application Layer

• Navigator: A graphical interface for creating interactive AI application workflows.
• Companion: A powerful chat interface for local and private LLM interaction.
• SDK / API: Python and Swift SDKs for programmatic interaction with the platform. This also includes our OpenAPI and OpenAI-compliant APIs.

Platform Layer

• Runtime: The orchestration layer that manages clusters and executes workloads designed in Navigator
• Network: The transport layer that provides peer-to-peer connectivity for distributed flows.

Software Requirements

Device Compatibility

webAI prioritizes delivering a seamless experience across various Apple devices. The platform is designed to ensure compatibility and optimized performance, whether you're using a cutting-edge iPhone or a high-powered Mac. Level of capabilities and performance will vary across devices.

Supported Devices: Mac computers with Apple Silicon (M1, M2, M3, M4, M5) are required.

Understanding AI Model Requirements

Different AI models have different resource needs. For most use cases that require training or inference on consumer hardware, we recommend using a 7B parameter model. This is the sweet spot between model performance and size for consumer devices.

Model Selection Factors

• Parameter Count: Generally, larger models offer better performance but require more computing resources
• Memory Requirements: Each model requires a specific amount of RAM
• Specialization: Some models are optimized for specific tasks (code generation, instruction following)

Choosing Models for Your Hardware

Recommended Model: 7B parameters

Smaller Models

• Good for basic conversational AI and simple tasks
• Run well on most supported devices
• Faster loading and response times

Medium Models

• Best balance of performance and hardware requirements
• Suitable for most professional use cases
• All LLM features in Navigator work well with these models

Larger Models

• Highest quality outputs and capabilities
• Require more powerful hardware
• Consider hardware capabilities before use

Performance Optimization

Hardware Considerations

• Apple Silicon Apple Silicon (M1, M2, M3, M4, M5) is required. 
• Available RAM: More memory allows for larger models and better performance
• Storage Type: SSD storage improves model loading times
• Free Space: Keep adequate space available for optimal performance

Improving Performance

System Optimization

• Close unnecessary applications during AI model use
• Ensure adequate free storage space
• Use SSD storage when possible
• Consider hardware upgrade for demanding use cases

Model Selection

• Start with smaller models to test performance
• Upgrade to larger models based on your hardware's capabilities
• Monitor system performance during use

Network Considerations

VPN Interference

Active VPN clients, especially those that route all traffic, can interfere with local network discovery and P2P communication between webAI components.

Corporate Networks

Corporate firewalls that block P2P traffic can prevent the Network layer from functioning correctly. Some restrictive NAT types (Symmetric NAT) may prevent direct connections between nodes.

Quick Compatibility Check

Before Installing

Check your Mac model and year

• Apple Menu -> About This Mac
• Note whether you have Apple Silicon (required) or Intel (not supported)

Verify macOS version

• Ensure macOS 15+ or later

Check available storage

• Ensure adequate space for installation and models

Test network connectivity

• Stable internet required for setup and model downloads

FAQ's

Does webAI support Windows or Linux?

At this time webAI supports the hardware outlined in this document. Check back for updates on additional hardware support.

What's Not Included (Out of Scope):

• Unsupported Hardware/Software: Any operating systems or hardware not explicitly listed in the "System Requirements" section. This includes, but is not limited to, Windows, Linux, and non-M-series Apple hardware, and jailbroken devices.
• Unsupported Network Configurations: WAN and IPv6
• Third-Party Software: Issues arising from software not provided by webAI
• Custom Code: Debugging or writing customer-specific code that interfaces with our SDKs/APIs beyond simple examples.
• End-of-Life Products: Support for versions of our products that are past their end-of-life date as defined in our support lifecycle.
• Performance Tuning & Optimization: While we will investigate performance-related bugs, we do not provide consultative performance tuning as part of our standard support.