GEX131 uses an RTX PRO 6000 Blackwell Max-Q (96 GB), outside the six-family comparison. Its prices load dynamically and were not readable (checked 14 October 2026); no rate is invented.

Checked 11 Oct 2026 · partial product record
Official product information ↗

Hyperstack

GPU virtual machines with published model-specific configurations.

Your part: You deploy, secure and operate the workload.
Routes: Rent a GPU machine
GPUs listed: H100 SXM · H100 PCIe · H200 SXM · A100 PCIe 80GB
Price basis: 4 current compute rates · checked 14 Oct 2026

Provider guide →Compare

View Hyperstack pricing

Billing notes & source

Good for
Run your own AI model on rented GPU infrastructure.

4 configuration records in this catalogue.

H200 and H100 SXM are listed as 8-GPU flavors. A per-GPU rate is not the full VM price.

Checked 11 Oct 2026 · partial product record
Official product information ↗

Lambda

Single- and multi-GPU instances, plus larger reserved clusters.

Your part: You deploy, secure and operate the workload.
Routes: Rent a GPU machine
GPUs listed: H100 SXM · H100 PCIe · B200 SXM 180GB · A100 SXM 80GB
Price basis: 6 current compute rates · checked 14 Oct 2026

Provider guide →Compare

View Lambda pricing

Billing notes & source

Good for
Run your own AI model on rented GPU infrastructure.

6 configuration records in this catalogue.

Single-GPU and 8-GPU plans have different per-GPU rates. Self-service stock is not checked.

Checked 11 Oct 2026 · partial product record
Official product information ↗

Modal

Serverless Python compute for inference, training, and batch jobs.

Your part: You manage code, data and scaling settings.
Routes: Run code on demand
GPUs listed: H100 SXM · H200 SXM · L40S PCIe · A100 (form factor unconfirmed)
Price basis: 4 current compute rates · checked 14 Oct 2026

Provider guide →Compare

View Modal pricing

Billing notes & source

Good for
Run AI code when you need compute, with platform scheduling.

4 configuration records in this catalogue.

GPU resource rates exclude separately metered CPU, memory, storage and paid plan charges. Image/video pipelines and agent code execution are documented capabilities (reviewed 12 October 2026). You still configure the model, workflow and agent tools; external model APIs can cost extra.

Checked 11 Oct 2026 · partial product record
Official product information ↗

Nebius

AI cloud with GPU VMs and cluster services.

Your part: APIs need application integration; GPU machines need workload operations.
Routes: Rent a GPU machine · Use a ready-to-call model
GPUs listed: H100 SXM · H200 SXM · B200 SXM 180GB · L40S PCIe
Price basis: 3 current compute rates · checked 14 Oct 2026

Provider guide →Compare

View Nebius pricing

Billing notes & source

Good for
Explore model APIs or GPU infrastructure, depending on the product.

3 configuration records in this catalogue.

Prices effective 1 October 2026 are used. Region-level capacity and account-specific quotas are not checked.

Checked 11 Oct 2026 · partial product record
Official product information ↗

Showing 6 of 12 matching providers

Operating guidance describes the product route, not a provider skill rating. Check the selected product’s documentation, current price and availability.

Continue a previous comparison

A virtual machine or serverless?

Compare DigitalOcean and Modal by setup, control and the way costs accrue.

Compare these routes →

Find a cloud.Know what comes next.

What do you want the cloud to do?

DigitalOcean

Hetzner

Hyperstack

Lambda

Modal

Nebius

A virtual machine or serverless?

Find a cloud.
Know what comes next.