GPU marketplace
Vast.ai
A marketplace of host-offered GPUs with competitive spot-style pricing for interruptible or carefully filtered capacity.
Product facts checked October 2, 2026Editorial scores last reviewed September 3, 2026
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.
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
Cost-sensitive batch jobs and experiments that can tolerate host variability
Watch out
Host inventory/reliability and offer duration vary. Interruptible instances may be paused when outbid or when on-demand capacity is requested. On-demand and reserved are high-priority and non-interruptible during the rental period, but this is an instance priority/resource guarantee, not a host-level uptime SLA. Reserved capacity is prepaid and may discount up to 50% based on commitment. Stopped-instance storage keeps billing until the instance is deleted.
| Billing | Per-second marketplace billing — interruptible, on-demand, or reserved |
|---|---|
| GPU choice | Wide, host-dependent inventory |
| Cold start | Instance provisioning + your stack |
| Model access | Self-managed models on rented machines |
Sources for the dated product and announcement claims
- vast-ai official product and billing documentation — source checked October 2, 2026
- vast-ai official product and billing documentation — source checked October 2, 2026
- vast-ai official product and billing documentation — source checked October 2, 2026
- vast-ai official product and billing documentation — source checked October 2, 2026
Fit detail
When Vast.ai is the right shortlist — and when it is not
Use these lists to kill bad comparisons early, before you compare logos.
Ideal for
- Price-hunting batch jobs and research sweeps
- Interruptible training or offline inference
- Teams willing to filter hosts carefully
Usually not ideal for
- Latency-critical user-facing chat without redundancy
- Compliance workloads that need a single accountable cloud
- Engineers who refuse to babysit host quality
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
Marketplace bids can undercut reserved clouds, but host variance and interruptions are part of the price. Budget engineering time, not just GPU hours.
Ops notes
- Check host reliability and the listed maximum rental duration before using on-demand for production.
- Use interruptible only for resumable work; save checkpoints because it can be paused.
- Delete stopped instances to end storage charges.
Fit scores
Vast.ai on the decision axes
Same editorial scale as the hub chart — useful for shortlists, not as a price quote.
Vast.ai
- GPU choice9/10
- Time-to-serving (editorial)5/10
- Price clarity7/10
- Production ops4/10
- Open-model breadth7/10
Among all 16 providers
GPU choice
Pick exact GPUs
Time-to-serving (editorial)
Warm-path fit
Price clarity
Easy to forecast
Production ops
Less DIY ops
Open-model breadth
Catalog depth
Scores are editorial planning ratings (1–10) for product shape — not published $/hour quotes or vendor SLAs. Verify current pricing on each provider site.
Compare
Vast.ai 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.