80% lower cold starts, 9 ms routing, and protected capacity at 2x the interruptible rate. Plus: switch URLs before August 1.
Lumenrack
Functions / June 2026

Real-time workloads.
Global footprint.

After months of work, four new features shipped in June to take real-time workloads global on Lumenrack Functions, our serverless container platform for CPU and GPU workloads.

80%
Lower cold starts
Checkpointing
9 ms
Routing decisions
Global routing
2x
Interruptible rate
Protected · July 1

01 · CPU and GPU checkpointing

Snapshot a fully warmed CPU and GPU container after model weights and kernels are loaded. New replicas restore from that snapshot instead of starting from scratch.

For vLLM, SGLang, and other GPU-heavy workloads, that turns multi-minute cold starts into seconds—and cuts wasted GPU time during scale-up.

Explore checkpointing →

02 · Global routing

Deploy to a single URL. Lumenrack routes every request to the closest healthy cluster across multiple regions and providers, making routing decisions in 9 ms. If a cluster degrades, traffic fails over automatically.

Explore global routing →

ACTION NEEDED · AUGUST 1

From August 1, 2026, legacy single-cluster URLs are no longer supported. They’ll keep working through a proxy, which can add latency. Switch to your global URL: it’s a host-only swap that takes about five minutes.

Migrate your URL →

03 · Global storage

Store once, read everywhere. One persistent storage layer is shared across all your apps and regions, so model weights and shared files no longer need to be duplicated per app or per region.

Explore global storage →

04 · Protected compute tiers

Choose a compute tier for each deployment.

Interruptible · default

Cheaper, preemptible capacity. Nothing changes for existing deployments.

Protected

On-demand capacity that Lumenrack does not reclaim while it is serving a request, for real-time and latency-sensitive workloads.

From July 1, 2026, protected is billed at 2x the interruptible rate. Interruptible pricing is unchanged. Enterprise customers with negotiated pricing are not affected.

Compare compute tiers →

From the Lumenrack blog

  • SOC 2 Type II
    We’ve completed our SOC 2 Type II audit and support GDPR, HIPAA, and ISO 27001 programmes.
  • Beacon, the Lumenrack global router
    How Beacon picks a healthy, cheaper, lower-latency cluster for every request in single-digit milliseconds, with automatic failover.
  • Restoring GPU workloads from memory snapshots
    How we snapshot warmed CPU and GPU memory inside our sandbox runtime and restore workloads like vLLM in seconds.

What would make Functions better for your workloads? Reply to this email—we’d love your feedback.

The Lumenrack team
Lumenrack on X Lumenrack on GitHub Lumenrack on LinkedIn Lumenrack community Slack
You’re receiving this because you have a Lumenrack Functions account.
Lumenrack, Inc., 1420 Halyard Row, Oakland, CA 94607
Copyright © 2026 Lumenrack, Inc. All rights reserved.
Privacy  ·  Terms  ·  Email preferences  ·  Unsubscribe