GPU list - Compute

compute gpu list
compute gpu list --json

That output is the source of truth for names and list ¢/hr. Marketing pages can lag.

v0.1 SKUs

Name	Also accepted	Provider	Role
`H100-SXM`	`H100`, `H100-80GB`	RunPod Secure	Public lead
`MI300X`	—	Hot Aisle	AMD; stock can be tight

Do not assume A100, consumer cards, community pools, or interruptible SKUs from this manual. If gpu list does not show it, do not pass it.

Decorator vs CLI

@app.function(gpu="H100-80GB", image=compute.Image.cuda_pytorch(), timeout=600)

compute run train.py::main --gpu H100-SXM

The CLI --gpu flag wins when both are set.

Images

NVIDIA: compute.Image.cuda_pytorch()
AMD: compute.Image.rocm_pytorch()

Use the image that matches the SKU. Mixing CUDA images onto MI300X (or the reverse) is a failed run, not a silent fallback. Rates: Pricing. Caps: Limits.

⌘I