Baseten GPU Pricing
Inference platform running custom and fine-tuned models on dedicated autoscaling GPUs with scale-to-zero, billed per minute with no idle charge; also sells per-token Model APIs.
GPU offers
6
Cheapest
$0.63/hr
T4
Billing
Per minute
Free credits
Credits for new accounts
Serverless
| GPU | VRAM | On-demand /hr | Spot /hr | Reserved /hr | Per month | Notes |
|---|---|---|---|---|---|---|
| B200 · compare all providers | ||||||
| B200 | 180GB | $9.98 | — | — | $7,285 | $0.16633/min dedicated deployment; includes the serving platform |
| H100 · compare all providers | ||||||
| H100 80GB | 80GB | $6.50 | — | — | $4,745 | $0.10833/min dedicated deployment; MIG half-H100 (40GB) $0.0625/min |
| A100 · compare all providers | ||||||
| A100 80GB | 80GB | $4.00 | — | — | $2,920 | $0.06667/min dedicated deployment |
| L4 · compare | ||||||
| L4 | 24GB | $0.85 | — | — | $619 | $0.01414/min dedicated deployment |
| A10 · compare | ||||||
| A10G | 24GB | $1.21 | — | — | $881 | $0.02012/min dedicated deployment |
| T4 · compare | ||||||
| T4 | 16GB | $0.63 | — | — | $461 | $0.01052/min dedicated deployment |
Per GPU-hour in USD, checked October 19, 2026. Monthly = 730 hours. Prices vary by region and availability; confirm on Baseten's pricing page.
Baseten GPU pricing FAQ
How much does B200 cost on Baseten?
Baseten lists B200 (B200) at $9.98 per GPU-hour on demand, about $7,285 per GPU per month running 24/7 (as of October 19, 2026).
How much does H100 cost on Baseten?
Baseten lists H100 (H100 80GB) at $6.50 per GPU-hour on demand, about $4,745 per GPU per month running 24/7 (as of October 19, 2026).
How much does A100 cost on Baseten?
Baseten lists A100 (A100 80GB) at $4.00 per GPU-hour on demand, about $2,920 per GPU per month running 24/7 (as of October 19, 2026).