AI Assistant Server Location

AI Assistant servers can be cloud-based, locally hosted, or connected via third-party endpoints, depending on configuration and provider.

Cloud-Based Servers

By default, AI Assistant accesses cloud-based models hosted by various providers such as OpenAI, Anthropic (Claude), Gemini, and Google Vertex AI. These servers are typically located in global data centers managed by the respective providers, and the exact region may vary depending on load balancing and service availability. For example, Gemini Code Assist operates globally through Google Cloud, and users cannot select a specific region for processing . Cloud servers provide instant access without requiring local infrastructure management .

Local Hosting

AI Assistant also supports locally hosted models or self-hosted MCP (Model Context Protocol) servers. In this setup, the server runs on your own machine or private infrastructure, giving you full control over data, security, and compliance. Local MCP servers communicate with AI Assistant via standard input/output or HTTP streaming, and can be configured using JSON files specifying the server executable and arguments . This is ideal for sensitive projects or offline use.

Third-Party Endpoints

You can configure AI Assistant to connect to third-party AI providers through API keys and endpoints. Supported providers include OpenAI-compatible services, Ollama, LM Studio, and others. These endpoints may be hosted in various regions depending on the provider, and the AI Assistant connects to them over the internet .

Geospatial and Specialized Servers

For AI applications requiring location-aware capabilities, services like the Mapbox MCP Server can be used. Mapbox offers both hosted servers for quick deployment and self-hosted versions for full infrastructure control, enabling AI agents to perform geocoding, route planning, and spatial queries .

Summary

• Cloud servers: Managed by providers, globally distributed, minimal setup.
• Local servers: Hosted on your own hardware, full control, offline capability.
• Third-party endpoints: Connect via API keys, region depends on provider.
• Specialized servers: Mapbox MCP Server for geospatial tasks, available hosted or self-hosted. Choosing between these options depends on your privacy requirements, latency needs, and integration complexity.