Hey Product Hunt 👋 I built Multiroute because every real product I work on ends up needing different models and different settings — reasoning effort, service tier, context shape — for different tasks. There was no clean way to manage that without scattering model-selection logic across the codebase. I wanted one self-configured endpoint with visibility into what was being routed where, that I could change without redeploying. What it does: - One OpenAI-compatible API across OpenAI, Anthropic, Google, Groq, Mistral, and more - Bring your own provider keys (BYOK) — keep your provider rates - Smart routing that picks service tier and reasoning effort per task, not just a model - Auto failover when a provider has a bad day - Optional LLM-as-judge evaluations to track quality across model swaps Pricing: $5 per 10,000 routed requests ($0.0005 each). New accounts get a $5 sign-up credit to try it out before topping up. Would love feedback from anyone running production AI workloads — especially on the routing tradeoffs you're making today and what you wish your gateway exposed. Thanks for checking it out!

Multiroute AI - smart LLM router

One API for every LLM — tuned per task, BYOK

One API for every LLM — tuned per task, BYOK