Service Level Agreement

The Commitment

Every GPU instance in the HygeiaCloud fleet carries a 99.5% monthly uptime commitment. Second-life hardware doesn't mean second-rate reliability. Every card in the fleet earned its place through a 72-hour burn-in and reference benchmarking, and we stand behind it with credits, not apologies.

How Uptime Is Measured

Uptime is measured per GPU instance over each calendar month, using our fleet monitoring. An instance counts as down for any minute it is unavailable or unable to execute workloads due to a failure of our hardware, network, or facility. Scheduled maintenance announced at least 72 hours in advance (and limited to 4 hours per month) is excluded from the calculation.

Service Credits

If monthly uptime for your instance falls below the commitment, you're entitled to a credit against that instance's monthly charges:

To claim a credit, email info@hygeiacloud.com within 30 days of the end of the affected month with the instance ID and dates. We check claims against our monitoring and apply approved credits to your next invoice.

What Happens When Hardware Fails

If a GPU degrades or fails, our monitoring flags it, we migrate your workload to a certified spare, and the failed card goes back through the full certification gauntlet before it can serve customers again. You should never be the one who discovers a hardware fault.

Exclusions

The commitment doesn't cover downtime caused by your software or configuration, suspension for breach of the service agreement, or events beyond our reasonable control (force majeure). This page summarizes our standard commitment. For customers with a signed service agreement, that agreement governs.

Questions about the SLA before you commit? Ask us directly. We'd rather you interrogate the fine print now than be surprised later.