These fall into two categories. Providers (which host GPUs and run models on them), and Routers (which handles picking an available provider for a model you want, and proxy the response).
This is the most popular LLM router.
OpenCode is an open-source coding harness for AI agents. You can use OpenCode with many different providers. But the people behind OpenCode also run two routers.
OpenCode Zen gives you some high-quality, tested and verified models with API pricing. They usually have some free models too. But essentially it’s pay-as-you-go.
OpenCode Go is a subscription service. You pay 10 dollars a month for access to open-weights models. You get different amounts of usage based on the model you pick, some of them are really cheap. And similar to other subscription LLM services, you get 5-hour, weekly and monthly rate limits.
You are able to use both of these services outside of OpenCode with a regular OpenAI-compatible API. I’ve heard of some people getting banned for “misusing” their Go subscription though.
Hvoy is not a router itself, but rather a website that provides uptime and pricing information for other Chinese LLM routers.