Pricing

Simple, transparent pricing for Phala Cloud compute and storage resources

Phala Cloud Pricing

We’ve made our pricing simple and clear so you can easily plan your costs.

How Pricing Works

• Pay-as-you-go: Only pay for time your CVMs are running
• Per-minute billing: Pay for exact minutes used
• No hidden fees: We track usage automatically
• No minimums: No subscription or minimum usage required

Compute and Storage

Pricing consists of two separate components:

Compute

• Billed based on the instance type you select
• Only charged while your CVM is running
• When you stop a CVM, compute billing stops immediately

Storage

• Billed at **$0.000139 per GB per hour** (~$0.10 per GB per month)
• Charged for both running and stopped CVMs
• To stop storage charges completely, you must delete the CVM
• Minimum disk size is 20GB per CVM
• System space is not included in disk allocation - you get the full disk size for your data

Pricing Example

A CVM using tdx.large (4 vCPU, 8GB RAM) with 100GB disk:

While running:

• Compute: $0.232/hour (based on instance type)
• Storage: 100 × $0.000139 =$0.0139/hour
• Total: ~$0.246/hour

While stopped:

• Compute: $0/hour
• Storage: 100 × $0.000139 =$0.0139/hour
• Total: ~$0.014/hour

Instance Types

We offer pre-configured instance types optimized for different workloads. View available instance types and pricing in the Phala Cloud dashboard.

Account Tiers and Limits

Tier	CVMs	vCPUs	Memory	Total Disk	Per-CVM Disk
Tier 1	8	16	32GB	160GB	80GB
Tier 2	16	32	64GB	1TB	256GB
Enterprise	Unlimited	Unlimited	Unlimited	Unlimited	Unlimited

New accounts start at Tier 1. Need more resources? Contact us to upgrade your tier.

GPU TEE Instances

GPU TEE instances provide confidential AI computing with hardware-level isolation for machine learning workloads.

GPU TEE Pricing

GPU Model	Commitment	Price	Notes
NVIDIA H200	On-Demand	$3.50/hour	24-hour minimum
NVIDIA H200	1-Month	$2.56/hour	Monthly commitment
NVIDIA H200	6-Month	$2.30/hour	Best value
NVIDIA B200	On-Demand	$3.80/hour	24-hour minimum

GPU TEE Billing Rules

GPU TEE uses hourly billing, unlike regular CVMs which bill per-minute.

On-Demand GPU TEE:

• Minimum charge: When you start an instance, you are charged $84 upfront for the first 24 hours (H200)
• When billing starts: No charges until the instance is actually launched
• Configuring Docker settings or topping up your account does NOT incur compute costs
• Usage beyond 24 hours is billed hourly (pay-as-you-go)
• If you stop and restart later, the 24-hour minimum applies again

Commitment Plans:

• Lower hourly rates with monthly or 6-month commitments
• Contact [e-mail] for commitment pricing

GPU Configuration

• Available GPUs per instance: 1, 2, 4, or 8 GPUs
• Multi-GPU pricing scales linearly (e.g., 2x H200 = 2x hourly rate)

Refunds

If GPU provisioning fails with “Resource matching failed: No teepod found matching all requirements”, contact [e-mail] with:

• Payment Order ID
• Order reference

Refunds are processed promptly for failed provisions.

What You Don’t Pay For

• Compute for stopped CVMs: $0 when powered off (storage fees still apply)
• Setup time: No charge for image pulling or creation
• Failed starts: Not charged if CVM fails to start

Network and Regions

• Regions: Currently available in US, with EU coming soon
• Pricing: Same rates across all regions (all prices in USD)
• Data Transfer: Each CVM includes bandwidth based on size

Legacy Free CVM Instance

For legacy accounts with existing free CVM instances:

• Legacy instances will be automatically reclaimed 6 months after their creation date
• Manual deletion may occur at any time for inactive or non-compliant instances
• Migration recommended to paid plans for continued service

Enterprise Support

For larger workloads:

• Volume discounts for sustained usage
• Annual billing options
• Custom support packages
• 99.9% uptime SLA

Questions? Contact us at [e-mail] for enterprise pricing.