Simple, transparent pricing

Per-second billing with no hidden fees. Hourly or monthly options.

Popular configurations

Starter points for common workloads.

Development

$0.44 / hr
RTX 4090 · 24 GB VRAM
  • Prototyping and local development
  • Instant SSH access
  • Jupyter support
Start

Training

$3.29 / hr
A100 · 40 GB VRAM
  • Model fine-tuning
  • Distributed workloads
  • Managed storage
Start

Production

$6.50 / hr
H100 · 80 GB HBM3
  • Large model training
  • Inference endpoints
  • High throughput workloads
Start

Serverless

Pay per request
Scale from zero
  • Sub-200ms cold starts
  • Zero idle billing
  • Autoscaling workers
Start

Enterprise

Custom pricing for reserved capacity, dedicated hosts, and support.

Reserved and dedicated capacity

Lower rates for committed use and dedicated host allocations.

Talk to Sales

Start with $10 in credits

No commitment. No lock-in. Pay by the second.

Get Started Read Docs