OrcaRouter LLM Router: Pricing, Features, Pros & Cons

OrcaRouter is an LLM router and AI gateway from Continuum AI, launched in May 2026. It exposes 200+ models through one OpenAI-compatible endpoint, and its orcarouter/auto mode grades each prompt and routes it to a model based on a quality and cost target you set, putting it alongside other LLM routers that choose a model per request rather than per app. Continuum AI reports that the adaptive router ranked second on the public RouterArena leaderboard at submission in May 2026.

Around the router, OrcaRouter adds failover chains, prompt caching, request logs, guardrails, an agent firewall that grades tool and MCP calls before they run, and versioned prompts. Logging of prompt and response bodies is off by default. There's no dataset or experiment-based evaluation in the product lineup, so measuring output quality over time needs a separate tool.

Pricing is built around zero markup: tokens bought through OrcaRouter bill at the provider's published rate, and the paid plans charge for team features like seats and compliance reporting. OrcaRouter Lite, an MIT-licensed edition, can be self-hosted with your own provider keys.