CoreWeave vs Spheron
Two GPU clouds, one question: who is actually cheaper for the GPU you need? Prices below are today's verified on-demand rates; platform facts come from each provider's own documentation.
Price per GPU, side by side (2026-07-08)
| GPU (on-demand $/hr) | CoreWeave | Spheron |
|---|---|---|
| RTX 5090 | not offered | $0.86 |
| RTX 4090 | not offered | $0.77 |
| B200 | $8.60 | $9.36 |
| H200 | $6.30 | $3.70 |
| GH200 | $6.50 | $1.88 |
| H100 | $6.16 | $2.01 |
| A100 | $2.70 | $1.43 |
| L40 | $1.25 | not offered |
| L40S | $2.25 | $1.89 |
| RTX PRO 6000 | $2.50 | $2.29 |
Cheapest verified on-demand rate per side from today's snapshot; highlighted = cheaper. Spot/marketplace tiers shown on the individual pages.
Platform facts
| CoreWeave | Spheron | |
|---|---|---|
| Billing increment | not published | not published |
| Egress | free | not published |
| Storage /GB-mo | $0.070/GB/month distributed file storage; $0.06/GB/month hot object storage | not published |
| Spot offered | not published | yes |
| Reserved offered | sales only | no |
| Notebook support | templates | templates |
From each provider's documentation, verified 2026-07-07; sources on the fees matrix and enterprise matrix.
Prices render from the live dataset each build. More: the screener · price movers · methodology.