Hi Product Hunt 👋 I built RelayModels because API bills were the main blocker for my own side projects: the same prompt could cost 5-10x more depending on where you buy the tokens, while the model itself is identical. So the whole product is about two things: 1. Aggregation. We connect a large number of upstream providers behind a single OpenAI-compatible endpoint, so pricing competition happens on our side, not in your budget. 39 models, one key, one base_url. 2. Verification. Cheap capacity is worthless if you silently get a downgraded model. Every route goes through automated checks — response fingerprinting, quality benchmarks, latency and error-rate monitoring. Anything that drifts is removed, and traffic fails over to a healthy provider automatically. You get the original models, per-token billing with no subscription, live usage logs (in/out/cached tokens per request), and reseller keys with quotas. Free $0.30 in credits after email confirmation — enough to run real tests. Would love your feedback on which models and features to add next.

RelayModels

Original AI models at a fraction of official API prices

Original AI models at a fraction of official API prices