OmniRoute

Overview

OmniRoute is a free and open-source AI gateway that presents many providers behind a single OpenAI-compatible endpoint at localhost:20128/v1. It provides automatic fallback, routing strategies, context compression, local-first storage, MCP and A2A exposure, and a local dashboard for managing providers, combos, health, costs, and analytics.

Key Features

• One OpenAI-compatible endpoint accepts OpenAI, Claude, Gemini, and Responses API requests at /v1.
• Automatic fallback and 19 routing strategies help keep requests moving when a provider is unavailable or out of quota.
• A 12-engine compression pipeline reduces eligible context by 15–95%.
• A local dashboard manages providers, combos, analytics, health, costs, quotas, and resilience controls.
• A built-in MCP server exposes 110 tools across 33 scopes, and an A2A server provides 6 skills over JSON-RPC 2.0.
• Self-hosted, local-first storage keeps operational data on the host, with encrypted credentials and optional proxy and TLS fingerprint controls.

Use Cases

• Use one base URL across coding tools such as Claude Code, Codex, Cursor, Cline, Copilot, and other OpenAI-compatible clients.
• Drain existing subscription or free-tier quota before routing to paid APIs.
• Reduce token costs with context compression and cost-optimized routing.
• Expose the gateway to MCP clients, A2A networks, or supported cloud coding agents.

Getting Started

• Install the omniroute npm package globally with npm install -g omniroute.
• Launch the gateway with omniroute; the dashboard and API serve on port 20128.
• Point an OpenAI-compatible client at http://localhost:20128/v1; the keyless auto route works on a fresh install.

Deployment & Requirements

• Deployment options include npm, Docker, a desktop app, ARM devices, Android/Termux, PWA, a VS Code extension, an OpenCode plugin, and building from source.
• The npm path installs the omniroute package globally; the Android/Termux path uses Node.js via pkg install nodejs and npx -y omniroute.
• Docker deployment uses a multi-arch AMD64 and ARM64 image.

Before You Adopt

• License: MIT. Review its terms before using, modifying, or distributing the project.
• The comparison is based on a public README and official documentation snapshot, so linked sources should be verified before purchase or deployment decisions.
• Availability depends on configured upstream providers; when a target is unavailable or out of quota, the gateway selects the next healthy target.
• In the self-hosted setup, credentials are stored in local SQLite with AES-256-GCM encryption.