No hidden fees. No egress charges. No surprise bills. Pay only for the GPU compute you use, with volume discounts that scale with your needs.
All prices are per GPU per hour. Volume discounts available for reserved capacity commitments.
| GPU Model | Memory | On-Demand | 1-Year Reserved | 3-Year Reserved | Spot |
|---|---|---|---|---|---|
| NVIDIA H100 SXM5 | 80GB HBM3 | $2.49/hr | $2.19/hr | $1.89/hr | $0.99/hr |
| NVIDIA H200 SXM | 141GB HBM3e | $3.89/hr | $3.49/hr | $2.99/hr | $1.59/hr |
| NVIDIA A100 SXM4 | 80GB HBM2e | $1.89/hr | $1.59/hr | $1.29/hr | $0.69/hr |
| NVIDIA L40S | 48GB GDDR6X | $1.29/hr | $1.09/hr | $0.89/hr | $0.49/hr |
* Spot pricing varies based on demand. Minimum 60-second billing increment. Egress is always free.
Every plan includes these essentials at no additional cost.
Unlimited data transfer out at no charge. No bandwidth metering, no surprise bills at month-end.
Data encrypted at rest and in transit by default. Customer-managed keys (BYOK) available on all plans.
GPU utilization, memory usage, and network metrics included. Prometheus-compatible metrics endpoint.
Full REST API, Python SDK, CLI tool, and Terraform provider included with every account.
Included high-performance storage with every cluster. Additional capacity at $0.02/GB/month.
Always-on network protection for inference endpoints. No additional cost, no configuration needed.
Everything you need to know about NeuralVane pricing and billing.
You're billed per GPU per hour with a minimum 60-second increment. Usage is metered in real-time and invoiced monthly. Reserved capacity is billed monthly regardless of usage, but at significantly reduced rates.
No. We don't charge for data egress, API calls, or networking between nodes in the same cluster. Storage beyond your plan's included amount is $0.02/GB/month. That's it.
On-demand gives you instant access with no commitment — pay only when you use GPUs. Reserved capacity commits to a 1 or 3-year term in exchange for 12-24% savings. You can mix both in the same account.
Spot instances offer up to 60% savings on unused GPU capacity. They can be preempted with 30 seconds notice when demand increases. Ideal for fault-tolerant training jobs with checkpointing enabled.
Yes. You can upgrade or downgrade your plan at any time. Changes take effect immediately. If you downgrade mid-cycle, you'll receive a prorated credit for the remainder of the billing period.
Yes. We offer special pricing for academic research institutions and early-stage startups (Series A or earlier). Contact our sales team for details on our NeuralVane for Startups program.
We accept all major credit cards, ACH bank transfers, and wire transfers. Enterprise customers can arrange custom payment terms including NET-30 and NET-60 invoicing.
No credit card required. Experience NeuralVane's full platform with complimentary compute credits.