GPU provider capabilities
Pause/resume, idle auto-stop, and persistent volumes aren't universal across GPU providers — this is what actually works on each one Aquanode supports, checked against the provider code itself rather than assumed from one provider that works. Snapshot and restore is the one row that's always yes: it's Aquanode's own mechanism, not the provider's, so your environment survives everywhere even where in-place pause doesn't.
| Provider | Pause / resume | Idle auto-stop | Persistent volumes | Snapshot / restore |
|---|---|---|---|---|
| DataCrunch | Yes | Yes | Yes | Yes |
| HyperStack | Yes | Yes | Yes | Yes |
| Voltage Park | Yes | Yes | Yes | Yes |
| RunPod | Partial | Yes | No | Yes |
| Vast.ai | Partial | Yes | No | Yes |
| SimplePod | Partial | Yes | No | Yes |
| Akash | No | No | No | Yes |
| Hotaisle | No | No | No | Yes |
| Vultr | No | No | No | Yes |
| Massed Compute | No | No | No | Yes |
“Partial” pause/resume means stop works but resume doesn't reliably — terminate and redeploy is the fallback there, same as on a “No” provider. Verified against each provider's adapter in Aquanode's orchestrator as of 2026-08-16.
What each column means
- Pause / resume
- Stop a running box in place and start the same box back up later, without redeploying. Where this is “No” or “Partial,” the environment is still not lost — see snapshot/restore below.
- Idle auto-stop
- Aquanode saves your environment and stops a box you left idle, so you stop paying for a GPU nobody's using. It only needs the provider's stop primitive, so it tracks the pause/resume column exactly.
- Persistent volumes
- Block storage you can attach, detach, and re-attach to a later deployment, independent of any one box's lifecycle.
- Snapshot / restore
- Save your whole environment and restore it onto a new box — on the same provider or a different one. This is Aquanode's own mechanism, so it works the same way everywhere; it's the fallback for “I have to rebuild my environment every time” on any provider above, pause/resume support or not. Automated snapshots are opt-in (15–30 min intervals in the console, as low as 1 min via the API/CLI) — nothing schedules one for you automatically, and snapshot storage isn't billed.
Per-provider notes
- Stop and resume both work end to end.
- Real block volumes: create, list, delete.
- Stop and resume both work end to end.
- Real block volumes: create, list, delete.
- Stop and resume both work end to end.
- Real persistent storage, but through Voltage Park's own separate storage API rather than the generic volume API most other adapters use (its CreateVolume method is intentionally unused).
- Stop works. Resume can come back with zero GPUs — a stopped pod stays pinned to its original machine, and if that GPU was taken while it was stopped, the pod restarts with none, is caught before being reported healthy, and has to be terminated and redeployed.
- Stop works. Resume does not — the client-level start call is an explicit "not yet implemented" stub, so a paused Vast.ai box cannot currently be resumed in place.
- Stop (pause) works. Resume does not — the underlying client method to resume a paused instance exists and works, but nothing in the adapter's start path calls it, so StartInstance always returns an error today.
- Neither stop nor resume is offered.
- Neither stop nor resume is offered.
- Neither stop nor resume is offered.
- Neither stop nor resume is offered.
Related pages
See live pricing per provider on the provider index, read more about idle auto-stop, network volumes, and workspace backups & migration, or browse every feature Aquanode ships.
