Skip to main content
← GPU inference hub

Serverless GPU

Modal

Serverless containers and GPUs billed for what you run — strong fit for Python inference jobs and bursty workloads without managing VMs.

Some outbound links use a first-party redirect hop for click counting. Commission is only claimed when a partner programme is active for that specific link — most vendor hops here are not paid placements. Affiliate disclosure.

Visit Modal(tracked)All providersSize VRAM firstEditorial scores last reviewed August 7, 2026

Marketplace referrals (for example RunPod or Vast) may return credits or kickbacks when a programme is active — that is a material connection even when it is not a cash CPA.

Best for

Developers who want GPU code as functions with autoscaling

Watch out

Cold starts and platform abstractions matter more than picking a bare metal SKU

BillingServerless CPU/GPU time
GPU choicePlatform-managed GPU classes
Cold startCan be noticeable on cold containers
Model accessPackage models in images or pull at runtime

Fit detail

When Modal is the right shortlist — and when it is not

Use these lists to kill bad comparisons early, before you compare logos.

Ideal for

  • Python inference jobs that spike and then go quiet
  • Teams that want functions + GPUs without VM babysitting
  • Pipelines that already look like jobs, not long-lived pods

Usually not ideal for

  • Needing one exact GPU SKU from a marketplace menu
  • Always-on low-latency endpoints where cold start is unacceptable
  • Non-Python stacks that fight the platform abstractions

Spend & ops

How cost behaves — and what breaks first

No invented $/hour quotes. These notes explain the failure modes that show up on the first real invoice.

Cost mental model

You mostly pay for what runs. Quiet hours are cheap; cold starts and oversized images show up as latency and surprise GPU-seconds.

Ops notes

  • Measure cold-start on the real image, not a hello-world container
  • Keep model weights close to the function region
  • Watch concurrency limits when traffic cliffs appear

Fit scores

Modal on the decision axes

Same editorial scale as the hub chart — useful for shortlists, not as a price quote.

  • GPU choice6/10
  • Time-to-serving (editorial)7/10
  • Price clarity7/10
  • Production ops8/10
  • Open-model breadth7/10

Compare

Modal vs alternatives

Open a head-to-head when you are deciding between product shapes.

Related

Keep the decision attached to the rest of the stack

Cloud rental is one path. Local VRAM and open-weight fit still matter.