How does Verlex make money?

The provider's GPU price passes through at cost. Verlex adds a fixed service fee of $0.05 to $0.75 per GPU-hour depending on the hardware tier and plan. No percentage markup, no hidden margin.

How much cheaper is Verlex than AWS or GCP list prices?

Every job is priced across 10 clouds and routed to the cheapest qualified GPU. Spot discounts at the provider typically run 60-80% off on-demand list. Example: an H100 at $3.50/hr list can run at about $1.05/hr spot plus a $0.40 service fee - about $1.45/hr, roughly 59% off all-in.

What happens if a spot GPU is reclaimed mid-job?

Checkpoints sync to object storage while the job runs. Verlex restarts the job from the last checkpoint on the next cheapest capacity, and the reclaimed time is not charged.

Can I get a surprise bill?

No. Credits are prepaid and jobs meter per second. When credits run out the job stops. Auto top-up is strictly opt-in, and every job has a hard 7-day backstop.

Do I need to change my code to run it on a cloud GPU?

No. results = verlex.cloud(train, gpu="A100") runs your existing Python function on a cloud GPU - two lines including the import.

Which clouds does Verlex route across?

AWS, Google Cloud, Azure, Verda, RunPod, Vast.ai, Hyperstack, JarvisLabs, TensorDock and Lyceum, plus serverless container lanes for small jobs on Beam, Northflank, Novita and Cerebrium.

Terms of Service

Effective date: July 3, 2026

1. Acceptance of Terms

By creating an account or using the Verlex platform ("the Service"), you agree to these Terms of Service ("Terms"). If you are using the Service on behalf of an organization, you represent that you have authority to bind that organization to these Terms. If you do not agree, do not use the Service.

2. Eligibility and Beta Status

The Service is currently in private beta. Access requires an invitation or beta key, and we are rolling out access in waves from our waitlist. During the beta, features may change, be limited, or be discontinued without notice. The Service is not available in jurisdictions where it would be prohibited by law.

3. Description of Service

Verlex is a cloud compute orchestration platform that routes Python workloads to low-cost available infrastructure across multiple cloud providers, including AWS, Google Cloud, Microsoft Azure, Verda, RunPod, Vast.ai, JarvisLabs, Hyperstack, TensorDock, Lyceum, Beam, Northflank, Novita, and Cerebrium. The Service may automatically fail over between providers, regions, and hardware types, and may run jobs on discounted spot capacity or in serverless container environments. We act as an intermediary, and we do not own the underlying compute infrastructure.

4. Acceptable Use

You agree not to use the Service to:

Execute malicious code, malware, ransomware, or spyware
Mine cryptocurrency without explicit authorization
Launch denial-of-service attacks or network intrusions
Violate the terms of service of any underlying cloud provider
Process, store, or transmit data that violates applicable privacy laws
Infringe on third-party intellectual property rights
Circumvent usage limits, billing systems, or access controls

Violation of these rules may result in immediate account suspension without refund.

5. Account Security

You are responsible for maintaining the confidentiality of your API keys and account credentials. We store API keys only as cryptographic hashes and cannot recover them for you; if a key is lost or compromised, revoke it and create a new one. You must notify us immediately at support@verlex.dev if you suspect unauthorized access. We are not liable for losses caused by unauthorized use of your credentials.

6. Credits and Billing

The Service operates on a prepaid credit model. Credits are purchased in advance and deducted as compute is consumed.

Compute is metered per second, with a minimum billing increment of one second
Rates are the underlying provider's cost passed through at cost, plus a fixed service fee per GPU-hour that depends on the hardware tier and your plan, as displayed in the product before you run a job
When we place a job on discounted spot capacity, you are billed at the spot rate actually obtained. If a spot instance is reclaimed by the provider, the reclaimed attempt is not charged and the job is rescheduled automatically
A credit hold may be placed on your balance while a job runs and is settled against actual usage when the job ends
Credits are non-refundable except as required by applicable law or at our sole discretion
Prices are quoted in USD. We support purchases in USD, BRL, CAD, INR, and THB, converted at current exchange rates. Prices may change with 30 days' written notice
If you enable automatic top-up, we will charge your saved payment method off-session for your configured top-up amount (default $50) whenever your balance falls below your configured threshold (default $10), in the same currency as your original purchase. You can change or disable automatic top-up at any time
We are not liable for failed jobs caused by third-party cloud provider outages. Credits consumed during such failures may be refunded at our discretion
All payments are processed by Stripe. By purchasing credits you agree to Stripe's terms of service

7. Performance Plan Subscription

In addition to prepaid credits, we offer an optional Performance plan as a recurring monthly subscription of $10 per month, billed through Stripe. The Performance plan unlocks fast execution mode, multi-GPU jobs on a single instance (up to 8 GPUs), and a larger storage quota. Performance-plan compute carries a higher fixed service fee per GPU-hour than Standard, as displayed in the product. You may cancel at any time through the billing page; your plan remains active until the end of the paid period, after which your account returns to the Standard plan. Subscription fees are separate from compute credits and do not add credit to your balance.

8. Job Execution Limits

Jobs run for as long as your credit balance can pay for them. A job is stopped when your credits are exhausted
Regardless of balance, every job is subject to an absolute maximum runtime backstop, currently 7 days, after which it is terminated
Jobs that show no compute activity for a sustained period may be stopped automatically to protect your balance
Spot-instance recovery is checkpoint-based: if your job writes checkpoints, we sync and restore them on the replacement instance; progress that was not checkpointed is lost and the job restarts from its last checkpoint or from scratch
Provisioning is subject to a time-bounded failover process across providers, regions, and hardware. If no capacity can be obtained within that window, the job fails rather than queueing indefinitely

9. Storage

Files you upload and outputs your jobs produce are stored in object storage under your account and count against your plan's storage quota (25 GB on Standard, 300 GB on Performance, larger quotas on Enterprise). Temporary workspace files are deleted automatically after a short period of inactivity, typically within about one day of last use. You are responsible for downloading or backing up outputs you need to keep long term.

10. Your Code and Data

You retain all intellectual property rights to the code and data you submit. By using the Service, you grant Verlex a limited, non-exclusive, royalty-free license to execute your workloads and store your files solely for the purpose of providing the Service. We do not inspect, train on, or sell your code or data. Compute instances are ephemeral and are destroyed or wiped after each job; job records, logs, and any outputs you store remain available in your account as described in our Privacy Policy.

11. Service Availability

The Service is in beta and we do not offer an uptime guarantee or service level agreement. We target high availability but do not guarantee uninterrupted or error-free service. Execution times depend on third-party cloud provider availability, which is outside our control. We are not liable for downtime, latency, or job failures caused by any underlying infrastructure provider.

12. Intellectual Property

The Verlex platform, including its software, design, trademarks, and documentation, is owned by Verlex and protected by applicable intellectual property laws. You may not copy, modify, distribute, or reverse-engineer any part of the platform without written permission.

13. Disclaimer of Warranties

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. WE DO NOT WARRANT THAT THE SERVICE WILL MEET YOUR REQUIREMENTS OR BE UNINTERRUPTED.

14. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, VERLEX SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES, INCLUDING LOSS OF DATA, PROFITS, OR BUSINESS, ARISING FROM YOUR USE OF OR INABILITY TO USE THE SERVICE. OUR TOTAL LIABILITY TO YOU FOR ANY CLAIM SHALL NOT EXCEED THE AMOUNT YOU PAID US IN THE 3 MONTHS PRECEDING THE CLAIM.

15. Indemnification

You agree to indemnify and hold Verlex harmless from any claims, damages, or expenses (including legal fees) arising from your use of the Service, your code or data, or your violation of these Terms.

16. Termination

We may suspend or terminate your account immediately for violations of these Terms, suspected fraud, or legal requirements. You may close your account at any time by contacting support@verlex.dev. Upon termination, unused credits are forfeited except as required by law.

17. Governing Law

These Terms are governed by the laws of the State of Delaware, United States, without regard to conflict of law principles. Any disputes shall be resolved exclusively in the courts of Delaware.

18. Changes to These Terms

We may update these Terms from time to time. We will notify you of material changes by email at least 14 days before they take effect. Continued use of the Service after the effective date constitutes acceptance of the updated Terms.

19. Contact

Questions about these Terms: support@verlex.dev

← Back to home