Description
Beam is a cloud platform for running GPU workloads, sandboxes and task queues for AI agents and applications. You can run custom model inference on serverless GPUs, start persistent Linux virtual machines, and create sandboxes that boot quickly and pause and resume with their state. The site describes sub-second cold starts, autoscaling, scale to zero, persistent local disks and a persistent IP address for each virtual machine, along with hardware isolation and secrets that never enter the machine. Beam can run on its own cloud or in your own AWS, GCP or bare metal environment. The pricing page lists sandboxes from $0.319 an hour, virtual machines from $0.0306 an hour and H100 GPUs from $1.83 an hour, with a calculator and options for clusters, storage and bring-your-own-cloud. The project's open-source code is on GitHub.…
Preview
Pricing
- SandboxesFrom $0.319/hr
- Virtual machinesFrom $0.0306/hr
- H100 GPUsFrom $1.83/hr
Highlights
- Serverless GPU inference
- Sandboxes that pause and resume
- Persistent Linux virtual machines
- Scale to zero and autoscaling
- Bring your own cloud
Ratings & Reviews
No ratings yet
Be the first to rate Beam.
People also visit
Beam FAQ
How much does Beam cost?
Pricing is pay as you go. The site lists sandboxes from $0.319 an hour, VMs from $0.0306 an hour and H100 GPUs from $1.83 an hour.
What can I run?
GPU inference, task queues, sandboxes and persistent virtual machines.
Can I use my own cloud?
Yes. The site lists bring-your-own-cloud on AWS, GCP or bare metal.
Is there an open-source part?
The site links to the beta9 project on GitHub.
What are the best Beam alternatives?
Popular alternatives to Beam include vMetal, TensorDock and paperspace. See all Beam alternatives









