What's good about...

Shadeform logo Shadeform

  • Launch instances anywhere with a single console and API
  • Leverage a multi-cloud set up, without the hassle
  • You can reserve instances in advance to ensure GPU availability
  • Free to use, zero mark-up fees

Shadeform logo TensorWave

  • Offers AMD GPUs as an alternative to Nvidia-based platforms
  • Native support for PyTorch and TensorFlow with no code changes
  • Can reserve for up to 3 years

Price comparison

Shadeform's Pricing

Shadeform does not charge any fees to use their marketplace. You pay the same price as using the cloud providers directly.

Here are some example configurations:

GPU VRAM Price range (per hour)
Nvidia A4000 16 GB $0.30-0.80
Nvidia A6000 48 GB $0.57-1.93
Nvidia V100 16 GB $0.69-3.06
Nvidia L40S 48 GB $0.99-1.15
Nvidia A100 40 GB $1.29-3.19
Nvidia H100 80 GB $2.48-5.99

You can view the full list here.

TensorWave's Pricing

TensorWave offers bare metal GPU nodes and fully managed Kubernetes clusters. You can reserve nodes for periods ranging from 6 months to 3 years.

While their pricing is not yet publicly available, you can request a quote on their website.

TensorWave offers GPUs in the following configurations:

Name GPUs VRAM vCPUs RAM Price/h
AMD MI300X 8x MI300X 1536GB -- -- On Request

Which services do they offer

Here are some managed services that Shadeform and TensorWave offer:

Service Shadeform TensorWave
GPU-powered Servers
Managed Containers --
Virtual Private Server (VPS) --

Company details

Shadeform
Website shadeform.ai
Headquarters United States of America πŸ‡ΊπŸ‡Έ
Founded 2023
Data Center Locations --
Example Customers University of California Berkeley, Magic Studio, Neets.ai, MK1
TensorWave
Website tensorwave.com
Headquarters United States of America πŸ‡ΊπŸ‡Έ
Founded 2023
Data Center Locations 2
Example Customers --

Alternatives to consider

Want to see how Shadeform and TensorWave compare against other providers? Check out these other comparisons:

More comparisons

Our data for Shadeform was last updated on Aug. 16, 2024, and for TensorWave on Oct. 6, 2024.