Network Virtualization for AI Infrastructure
Kapapack is a high-performance network simulator designed for modern AI datacenters. Test and optimize your distributed training workloads before deploying to production hardware.
Simulate complex network topologies, custom congestion control algorithms, and collective communication patterns with ease. Our Python API integrates seamlessly with PyTorch and other ML frameworks.
Try Our Interactive Demo
Experience Kapapack's capabilities firsthand.
Configure
workloads, test algorithms, and simulate datacenter networks.
Fast & Lightweight
Built for speed. Simulate large-scale datacenter networks with
minimal overhead.
PyTorch Integration
Native support for NCCL and distributed training workflows with
PyTorch.
Easy Python API
Intuitive API for defining topologies, algorithms, and workloads.