Inferock

Not yet independently verified. The launch date rests on the maker's Product Hunt launch of 14 October 2026. The encryption and service-credit terms are the maker's own descriptions of its intended design and were not independently checked. We will update this when it can be confirmed, and remove this note.

Inferock is a service for running AI model inference with a record of what each call did. It offers three ways to use it. Inferock Watch is free and sits in front of a customer's own provider accounts, recording quality findings, latency, usage and cost for each call, with a monthly performance report; the providers bill the customer directly. Inferock Serve is prepaid serverless inference with no provider key needed: per-token model rates include the provider's usage, and the maker states a 10 percent service fee on actual supplier cost, with non-cash service credits for eligible failures under agreed terms. Inferock Dedicated is reserved capacity priced by a quote agreed before anything is provisioned. The maker also describes an intended confidential route, designed to encrypt content end to end between the client and a verified workload, and says each route must be qualified with evidence for its exact model and configuration; it states that this protection does not apply to Watch connections. It publishes an open-source benchmarking tool, inferock-bench. Inferock launched on Product Hunt on 14 October 2026; its site does not state where the company is based.