Hi — I'm Arun, founder of Mua Technologies. I built Throughputs because the bill was the problem, not the models. The short version: you keep your existing code. Throughputs is an OpenAI-compatible endpoint, so in most apps the change is one line — the base URL — and nothing else. Underneath, it serves open-weight and frontier models behind that one interface, and picks healthy capacity per request rather than paying for a fixed always-on region. Who it's for: teams shipping LLM features who watch the inference line item at the end of the month. Who it's not for: anyone wanting a chat app — this is infrastructure, you'll be in a terminal. We're a small team in Chennai, and I read everything here. The most useful thing you can tell me is where it broke for you — setup friction, model gaps, docs that assumed the wrong thing. Those become next week's commits. Thanks for taking a look.

Throughputs

A affordable server-less inferencing engine that serves llms

A affordable server-less inferencing engine that serves llms