Inference is the operational half of machine learning and the half that costs money every day. Training is a capital expense; inference is the running cost, and optimising it is where a great deal of current engineering effort goes.
This name is the vendor-neutral version of the term, which matters. It covers every model family rather than tying the brand to one provider’s naming, which makes it more durable as the market shifts between providers.
It fits an inference service, an optimisation product, a hardware company, a benchmarking resource, or a cost management tool. The hardware reading is worth noting: inference-specific silicon is a serious and well-funded category with several credible entrants.
Twelve characters, two components, no hyphen, .com, and both are standard vocabulary. The double i in the middle is the one thing to note, and it reads cleanly enough in lowercase that it causes no real difficulty.
Enquiries welcome.