The verdict
Vast.ai takes it, 4–2 on rounds.
On the published figures and the independent review sites, Vast.ai takes 4 of 6 rounds. Treat that as a starting point and check the current figures on both pages, because prices and plans move.
Choose Vast.ai if…
- • You want Vast.ai's edge on the cheapest h100 and the cheapest a100
Choose CoreWeave if…
- • You want CoreWeave's edge on what it costs to get your data out and how much you save by committing
Round by round
- Round 1The cheapest H100Vast.ai winsVast.ai8/10
$1.48
H100 SXM lowest price
- the lowest H100 offer on the marketplace (offers seen ran from about $2 to $19)
Source: vast.ai ↗
CoreWeave5/10$6.16
per GPU, worked out from $49.24 an hour for an HGX H100 with eight GPUs
- $6.16, against $0.90
Source: coreweave.com ↗
Vast.ai wins this round on what the company itself publishes.
- Round 2The cheapest A100Vast.ai winsVast.ai8/10
$0.37/hr
lowest A100 PCIe offer
- the lowest A100 PCIe offer on the marketplace
Source: vast.ai ↗
CoreWeave5/10$2.70
per GPU, worked out from $21.60 an hour for an eight-A100 instance
- $2.70, against $0.35
Source: coreweave.com ↗
Vast.ai wins this round on what the company itself publishes.
- Round 3How cheap you can startVast.ai winsVast.ai8/10
$0.11
RTX 3090 hourly rate
- the lowest rate on the marketplace, for consumer GPUs
Source: vast.ai ↗
CoreWeave5/10$1.25
the L40, worked out from $10.00 an hour for eight
- $1.25, against $0.13
Source: coreweave.com ↗
Vast.ai wins this round on what the company itself publishes.
- Round 4What it costs to get your data outCoreWeave winsVast.ai5/10
Bandwidth charged both ways
billed per byte sent or received, at rates that vary by host
- Bandwidth charged both ways, against Free
Source: docs.vast.ai ↗
CoreWeave8/10Free
data transfer between CoreWeave and the internet is listed as free
Source: coreweave.com ↗
CoreWeave wins this round on what the company itself publishes.
- Round 5How many places it runsVast.ai winsVast.ai8/10
40+ data centers
Marketplace coverage wording: 'Prices set by supply and demand across 40+ data centers.' Homepage also says 'On-demand instances across 40+ data centers and 20,000+ GPUs.'
- host locations on a marketplace, not data-center regions
Source: vast.ai ↗
CoreWeave5/1051 data centers
operated across North America and Europe, counted as data centers rather than regions
- 51 data centers, against 500+ locations
Source: coreweave.com ↗
Vast.ai wins this round on what the company itself publishes.
- Round 6How much you save by committingCoreWeave winsVast.ai5/10
Up to 50% off
marketplace GPU cloud (on-demand / interruptible / reserved)
- Up to 50% off, against Up to 60% off
Source: vast.ai ↗
CoreWeave wins this round on what the published sources say.
Side by side
- the lowest H100 offer on the marketplace (offers seen ran from about $2 to $19)
- the lowest A100 PCIe offer on the marketplace
- the lowest rate on the marketplace, for consumer GPUs
- host locations on a marketplace, not data-center regions
- per GPU, worked out from $21.60 an hour for an eight-A100 instance
- on-demand GPU instances, North America
- Bandwidth charged both ways, against No egress fees
- Bandwidth charged both ways, against Free
- Up to 50% off, against Up to 60% off
- $0.13, against Free GPU notebooks
- $6.16, against $1.99
- $2.70, against $1.19
- $6.31, against $3.59
- $8.60, against $5.98