Overview

Principles

OneRouter is built on four core principles that guide every architectural decision. Understanding these helps you build better, more resilient AI applications.

1. Zero Provider Lock-In

Every AI provider has strengths and weaknesses. Models improve weekly. Prices change constantly. OneRouter ensures you're never trapped with a single vendor.

Switch models with a single parameter change. No SDK rewrites. No pipeline rebuilds. Your code calls gpt-4o today and claude-opus-4-8 tomorrow with zero friction.

Pro tip: Use auto-failover to automatically fall back to a secondary model when your primary provider experiences downtime. Zero dropped requests.

2. Transparent, Pay-As-You-Go Pricing

OneRouter charges a small, transparent service fee on top of upstream provider rates — no hidden costs, no surprise bills. You always see the exact per-token price before making a call.

  • Every model's per-token cost is visible in real-time on the dashboard
  • No minimum commitment — top up any amount and start building
  • Set per-key budgets and spending caps for full cost control
  • Pay only for the tokens you consume — no monthly subscription required

3. Global Low-Latency Access

OneRouter routes through edge infrastructure worldwide, aiming for sub-50ms routing overhead for developers globally. Self-hosted deployments can be placed in any region for optimal latency to your users.

Our infrastructure is designed for resilience: horizontal scaling, automatic failover, and no single point of failure.

4. Privacy-First Architecture

OneRouter does not log or store your prompt/response content by default. Your data passes through our infrastructure in real-time. For self-hosted deployments, you have full control over logging policies. API keys are encrypted at rest (AES-256) and all traffic uses TLS 1.3.

For enterprises with strict data residency requirements, OneRouter is fully self-hostable via Docker or Kubernetes — deploy behind your firewall with complete data sovereignty.