modelsplit.com
ModelSplit.com could define the infrastructure layer for partitioning AI models across GPUs, edge devices and distributed compute — placing each portion of a model where latency, memory, cost and available hardware make the most sense.