Grant-free multiple access (GFMA) is an emerging technology to accommodate a massive number of devices for 6G-enabled Internet of Things (IoT) networks. The main advantages of GFMA are to efficiently reduce control signaling overhead for resource scheduling while improving resource efficiency. In this article, we propose a novel resource-hopping-based GFMA (RH-GFMA) framework with resource hopping schemes for providing massive connectivity in 6G cellular IoT networks, where each IoT device is allowed to access physical radio resources by using a preassigned resource hopping pattern without not only resource request but also grant procedure, which is the so-called "one-shot" noninteractive multiple access. We exploit three types of resource hopping schemes in the proposed RH-GFMA framework: 1) random hopping; 2) resource group hopping; and 3) Latin-square group hopping. We mathematically analyze the RH-GFMA system performance in terms of the hopping pattern collision probability, maximum allowable packet delay, and interference-over-thermal. Finally, we derive an accommodation capacity of the proposed RH-GFMA framework, which is defined as the expected number of IoT devices accommodated in a cell under a maximum allowable packet-delay requirement and an interference-over-thermal constraint. With the proposed GFMA, massive IoT devices are expected to be efficiently accommodated in 6G wireless networks, while satisfying strict latency and reliability requirements.