Execute untrusted code, parallel browser sandboxes, and autonomous AI agents inside ephemeral Firecracker MicroVMs spawned on bare-metal silicon. Zero cold-start latency. Zero shared state.
Click any stage in the distributed pipeline to inspect real hardware telemetry.
BGP Anycast routing terminates TLS 1.3 with zero round-trip overhead. Ingress requests are verified and serialized within 800 microseconds.
Adjust parallel worker concurrency to evaluate latency degradation curve.
from hyperplane import HyperplaneClient
# Connect to low-latency kernel gateway
client = HyperplaneClient(token="hp_live_9f81a7b2")
# Spawn an isolated 4.8ms MicroVM instance
vm = client.sandboxes.spawn(
image="python:3.12-alpine",
cpus=4,
memory_mb=512,
isolation="firecracker-kvm"
)
# Run arbitrary deterministic agent workload
result = vm.exec(
"pytest tests/test_migrations.py --benchmark-json output.json"
)
print(f"Boot: {vm.boot_latency_ms}ms | Status: {result.exit_code}")
print(result.stdout)