CoreWeave vs Vast.ai
SEPT 2026 auditA comparison of CoreWeave and Vast.ai built from values read directly from each provider's own documentation, with the source recorded against every figure.
The short answer
Choose CoreWeave if…
Training runs that want a whole HGX node and move large datasets — free egress and free I/O operations remove the charge that usually dominates a data-heavy job.
Editorial · Palash Bagchi · approved
Choose Vast.ai if…
Price-sensitive training and batch inference that can tolerate a marketplace — the median H100 rate is well under half what the managed clouds charge, and interruptible capacity goes lower still.
Editorial · Palash Bagchi · approved
Consider something else if…
- CoreWeave: You cannot rent a single card: the H100 is sold as an eight-GPU node at $49.24 an hour, so a one-GPU experiment costs eight GPUs.
- Vast.ai: There is no single price to plan against: rates move with supply, the host is not Vast itself, and an interruptible instance can be reclaimed mid-run.
3 sourced criteria separate them — see where, with sources, below.
Pick the criteria you care about. The chart counts how many of them lean toward each provider — the same read as scanning the bars below, just totalled for the ones you chose.
CoreWeave 0
Vast.ai 0
At a glance.
| Criterion | CoreWeave | Vast.ai |
|---|---|---|
| GPU | ||
| H100 SXM, per GPU-hour | $6.16 per GPU-hour for H100 SXM 80GB. CoreWeave sells the 8-GPU HGX H100 node at $49.24/hour on demand — the per-GPU figure is that divided by eight, ours rather than CoreWeave's, and you cannot rent one card. | Median $2.16 per hour for H100 SXM 80GB, with a floor of $1.33 from the cheapest host. Vast is a marketplace — the floor is one host's offer at one moment and the median is what the market is actually charging, so both move without anyone announcing anything. This figure is a snapshot, not a rate card. |
| Cheapest GPU, per hour | $1.25 per GPU-hour for L40, again as an 8-GPU node at $10.00/hour. A100 80GB nodes are $21.60/hour, or $2.70 per GPU. | Floor and median vary by card and by the hour; H100 PCIe runs from $1.87 with a median of $2.67, and H200 from $1.97 with a median of $4.24. The H200 floor now sits below the H100 PCIe median, which is the kind of inversion a marketplace produces and a rate card cannot. |
| Largest GPU offered | 288GB per GPU on B300-class hardware; the largest listed on-demand node is HGX B200 at 180GB per GPU, eight GPUs, $68.80/hour. | 192GB per GPU on B200, from $4.38 an hour with a median of $7.50. |
Where they differ.
H100 SXM, per GPU-hour
- CoreWeave
- $6.16 per GPU-hour for H100 SXM 80GB. CoreWeave sells the 8-GPU HGX H100 node at $49.24/hour on demand — the per-GPU figure is that divided by eight, ours rather than CoreWeave's, and you cannot rent one card.
- Vast.ai
- Median $2.16 per hour for H100 SXM 80GB, with a floor of $1.33 from the cheapest host. Vast is a marketplace — the floor is one host's offer at one moment and the median is what the market is actually charging, so both move without anyone announcing anything. This figure is a snapshot, not a rate card.
Sources (2) →Sources ↓
- Pricing — CoreWeave ↗
“NVIDIA HGX H100 8 80 128 2,048 61.44 $49.24 NVIDIA HGX H100 On-Demand”
Read 2026-09-06 · official pricing
- Pricing — Vast.ai ↗
“H100 SXM Hopper 80GB VRAM from $1.33 /hr median $2.16/hr”
Read 2026-09-08 · official pricing
Cheapest GPU, per hour
- CoreWeave
- $1.25 per GPU-hour for L40, again as an 8-GPU node at $10.00/hour. A100 80GB nodes are $21.60/hour, or $2.70 per GPU.
- Vast.ai
- Floor and median vary by card and by the hour; H100 PCIe runs from $1.87 with a median of $2.67, and H200 from $1.97 with a median of $4.24. The H200 floor now sits below the H100 PCIe median, which is the kind of inversion a marketplace produces and a rate card cannot.
Sources (2) →Sources ↓
- Pricing — CoreWeave ↗
“NVIDIA L40 8 48 128 1,024 7.68 $10.00 [...] NVIDIA A100 8 80 128 2,048 7.68 $21.60”
Read 2026-09-06 · official pricing
- Pricing — Vast.ai ↗
“H100 PCIE Hopper 80GB VRAM from $1.87 /hr median $2.67/hr [...] H200 Hopper 141GB VRAM from $1.97 /hr median $4.24/hr”
Read 2026-09-08 · official pricing
Largest GPU offered
288GB per GPU on B300-class hardware; the largest listed on-demand node is HGX B200 at 180GB per GPU, eight GPUs, $68.80/hour.
192GB per GPU on B200, from $4.38 an hour with a median of $7.50.
Sources (2) →Sources ↓
- Pricing — CoreWeave ↗
“NVIDIA HGX B200 8 180 128 2,048 61.44 $68.80”
Read 2026-09-06 · official pricing
- Pricing — Vast.ai ↗
“B200 Blackwell 192GB VRAM from $4.38 /hr median $7.50/hr”
Read 2026-09-08 · official pricing
When these numbers change, hear about it. Sources are re-checked monthly; a repricing goes out as a short note.
Documented by only one.
These criteria are published by one provider and not the other. An absence here means we have not found a source, not that the feature is missing.
Questions this comparison answers.
Should I choose CoreWeave or Vast.ai?
Pick CoreWeave if Training runs that want a whole HGX node and move large datasets — free egress and free I/O operations remove the charge that usually dominates a data-heavy job.
Pick Vast.ai if Price-sensitive training and batch inference that can tolerate a marketplace — the median H100 rate is well under half what the managed clouds charge, and interruptible capacity goes lower still.