Hi Product Hunt. I'm George, and I built IQ Routing solo. The idea started from a problem I kept seeing in my own AI-agent usage: agent runs can get expensive fast, and a lot of that spend is hard to justify. Most routing tools treat an agent run like a pile of separate requests. They look at each request on its own and try to make that single request cheaper or faster. But that misses how agents actually work. An agent run is one flow from start to finish. Some steps are busywork. Some steps are context-building. A few steps are the decisions that determine whether the final result is good or bad. Paying for the strongest model on every step is wasteful. Using a weaker model everywhere is risky. The right answer is to spend more only where the work actually needs it. That is what IQ Routing does. It sits between your agent tools and model providers, looks at each step in the run, and picks the cheapest model that can still do that step well. It also reuses work it has seen before, so you are not paying repeatedly for answers the system already knows. The setup is meant to be boring in a good way: point your existing tool at IQ, keep your code the same, and keep working. IQ works today with Claude Code, Codex, and other popular apps This is the launch. Try it, break it, and tell me where the routing feels wrong. I read every reply.

IQ Routing

Trajectory-aware LLM routing that cuts agent cost

Trajectory-aware LLM routing that cuts agent cost