Does Hyperstack have a free tier?

No, Hyperstack does not currently offer a free tier.

Is Hyperstack open source?

No, Hyperstack is not open source.

Hyperstack

On-demand cloud GPU platform for AI and ML workloads with per-minute billing

Hyperstack by NexGen Cloud provides on-demand access to NVIDIA GPUs (H200, H100, A100, L40) for AI and ML workloads. Instances deploy in minutes with per-minute billing. Also offers AI Studio for building and deploying models, managed Kubernetes, and developer SDKs for Python, Go, and TypeScript. Data centers run on renewable energy across Europe and North America.

Pricing: Hourly

Hosting Cloud

Pricing Usage Based, from $0.15/hr

HQ 🇬🇧 United Kingdom

Founded 2020

License PROPRIETARY

Compliance SOC 2

Visit website →

Pricing

Posts

What Hyperstack is

Hyperstack is an on-demand GPU cloud from NexGen Cloud, aimed at AI and ML training and inference workloads. Instances deploy in minutes and are billed per minute, so short jobs do not pay for a full hour. Data centers run on renewable energy across Europe and North America.

GPUs and pricing

On-demand rates (per their pricing page, June 2026) span from $0.15/hr for an A4000 (16GB) up to $3.50/hr for an H200 SXM (141GB). Common mid-range options include the L40 (48GB) at $1.00/hr, the A100 (80GB) from $1.35/hr, and the H100 (80GB) from $1.90/hr. Reserved instances cut the hourly rate (for example H100 from $1.33/hr) in exchange for commitment, and spot VMs are available on a subset of GPUs at lower prices. There is no free tier.

Beyond raw GPUs

Hyperstack also offers AI Studio for building and deploying models, managed Kubernetes (the master node is free), and SDKs for Python, Go, and TypeScript. Data transfer is not charged, and persistent storage is metered at roughly $0.10/TB/hour. SOC 2 certified.

Who it fits

It suits teams that want straightforward per-minute GPU rental with predictable on-demand rates and the option to reserve for steady workloads. Compared to a marketplace like Vast.ai, where prices float with supply and demand, Hyperstack publishes fixed rate cards, which trades the chance of marketplace bargains for predictable billing. Compared to hyperscaler GPU instances, the per-hour rates are generally lower, with the trade-off of a smaller region footprint.