GPU.ai
vs
Thunder Compute
Both rent A100, A6000, H100 and L40 GPUs. GPU.ai routes each launch to the cheapest upstream cloud, while Thunder Compute virtualizes GPUs over the network and bills per minute.
What's good about each
- Per-second billing with no minimum commitment
- Aggregates capacity across 14+ upstream providers on three continents
- Public pricing API exposing live per-region availability counts
- No egress or networking fees
- Per-minute billing, nothing charged while an instance is stopped
- Per-second GPU sandboxes, each its own Firecracker microVM
- MCP server for Claude Code, Cursor and Codex
- 100 GB persistent disk per GPU, no egress fees
Price comparison
Closest matching configuration on each side, monthly list price in USD.
| Service | GPU.ai | Thunder Compute | Difference |
|---|---|---|---|
| Block storage | -- Not listed | $21.90 /mo 100 GB beyond free allowance · First 100 GB per GPU included | |
| Egress | -- Not listed | Free and unlimited |
Heads up: Figures take each provider's closest plan to a common setup, and vary by region and usage. Hover a service for the reference configuration. Verify before provisioning. More on how we price.
GPUs
GPU.ai offers 20 GPU models, while Thunder Compute offers 4. Cheapest listed price per GPU per hour; GPUs both providers sell come first.
| GPU | GPU.ai | Thunder Compute | Difference |
|---|---|---|---|
H100
8 configs · 4 configs
|
$1.98 /GPU/hr On-demand | $3.20 /GPU/hr On-demand · sold out | 38% cheaper on GPU.ai |
|
A100
4 configs · 4 configs
|
$1.07 /GPU/hr On-demand | $1.09 /GPU/hr On-demand | Within 5% |
|
RTX A6000
4 configs · 1 config
|
$0.40 /GPU/hr On-demand | $0.35 /GPU/hr On-demand | 13% cheaper on Thunder Compute |
|
L40
3 configs · 4 configs
|
$0.78 /GPU/hr On-demand | $0.79 /GPU/hr On-demand · sold out | Within 5% |
|
B300
2 configs
|
$8.99 /GPU/hr On-demand | -- | GPU.ai only |
|
B200
1 config
|
$7.99 /GPU/hr On-demand | -- | GPU.ai only |
|
RTX PRO 6000
4 configs
|
$1.69 /GPU/hr On-demand | -- | GPU.ai only |
|
H200
3 configs
|
$3.79 /GPU/hr On-demand | -- | GPU.ai only |
|
RTX 5090
1 config
|
$1.19 /GPU/hr On-demand | -- | GPU.ai only |
|
RTX 4090
1 config
|
$0.34 /GPU/hr On-demand | -- | GPU.ai only |
GPU.ai bills on-demand; Thunder Compute bills on-demand. The line under each price names the billing type shown.
Services
Managed services each provider offers.
| Service | GPU.ai | Thunder Compute |
|---|---|---|
| Block Storage | ||
| GPU-powered Servers |
Compliance
Frameworks each provider states on its own site, each linked to its source.
Heads up: Each entry links to the provider's own statement. An attestation can cover one tier, one region or the platform alone rather than every service. Verify the scope before you rely on it. A dash means we found nothing to link to, not that the provider lacks it.
Company details
| Fact | GPU.ai | Thunder Compute |
|---|---|---|
| Founded | -- | 2024 |
| Headquarters | UAE | USA |
| Locations | -- | -- |
| Example customers | -- | NYU Langone Health, MIT, ETH Zürich, UC Berkeley, Theseus, Conduit, InstaLILY |
| Website | gpu.ai | www.thundercompute.com |
Alternatives to consider
Compare GPU.ai and Thunder Compute against other cloud providers.
Our data for GPU.ai was last updated on Oct. 21, 2026, and for Thunder Compute on Oct. 21, 2026.