OmniRoute

OmniRoute is an AI gateway for developers who want their coding assistants and other AI clients to share one set of provider connections. It routes requests across models and accounts when a service fails, reaches a rate limit or runs out of quota.

• Routing and fallback: Choose routes by cost, available quota, context size or load, and build model groups that switch automatically to another healthy target.
• Client and model compatibility: Connect Claude Code, Codex, Cursor and Cline to cloud providers such as OpenAI and Anthropic, or local runtimes including Ollama, LM Studio, vLLM and llama.cpp.
• Context and caching: Compress eligible context and tool output before sending it to a model, reuse cached responses, and retain conversational memory with keyword and vector recall.
• Dashboard and controls: Manage providers and routing rules, inspect usage and latency, and apply prompt-injection checks or optional personal-data redaction from a local dashboard.
• Agent connections: Expose gateway functions through MCP and A2A, and connect Notion or Obsidian as context sources through MCP.

OmniRoute is free, open source and MIT-licensed. It runs on Windows, macOS and Linux, supports Docker and ARM hardware including Apple Silicon and Raspberry Pi, and can run on Android through Termux. No OmniRoute account is required. Credentials, usage records and history stay on your host, with encrypted credential storage; requests routed to external providers leave that host for inference. Cheaper Inference is a separate hosted gateway offering cost-ranked provider routing.