Pay only for what you use

Granular per-second billing with no idle costs. No contracts, no commitments, no surprises.

Per-second GPU pricing

All plans include autoscaling, global deployment, observability, and 99.9% uptime SLA. Volume discounts available for committed usage.

$0.0012/ per second

Cost-effective for smaller models and batch processing

$0.0042/ per second

Production-grade for most AI workloads

$0.0088/ per second

Latest hardware for large language models

You're billed for the exact GPU time used during inference, rounded to the nearest second. No charges when your endpoint is idle or scaling to zero.

Yes. Committed usage of $5,000/month or more qualifies for volume pricing. Contact our sales team for a custom quote.

GPU compute, networking, storage, observability tools, autoscaling, and support. No hidden fees or add-ons.

Yes. Change GPU types at any time by updating your deployment config. Billing adjusts automatically to the new rate.

For large-scale deployments we offer reserved capacity contracts with discounted rates. This guarantees GPU availability during high-demand periods.

You can set spending limits and alerts in the dashboard. When a limit is reached, we'll notify you and optionally pause deployments.

New accounts receive $100 in GPU credits. No credit card required to start.