Verlex

Verlex is a platform designed to simplify and automate access to GPU compute resources across multiple cloud providers. It enables users to run Python functions on the most cost-effective GPU hardware available at the time of execution, without manual intervention in hardware selection, price comparison, or deployment scripting. ai, Hyperstack, JarvisLabs, TensorDock, and Lyceum—and executes the job on the cheapest available option.

The platform features automatic failover, ensuring that if a provider is out of stock or fails, the job is automatically restarted on the next cheapest provider. Verlex handles provisioning, driver installation, dependency management, checkpoint recovery for spot instances, and metering. Users are shown the price before each job runs, and billing is calculated per second, with a one-second minimum increment. There are no quotas, consoles, or deployment scripts required, and the system is designed to eliminate the need for manual DevOps processes.

Verlex also supports hybrid execution, allowing local machines to offload heavy workloads to the cloud only when resource thresholds are exceeded. This is managed automatically, with no need for annotations or cluster setup, and ensures that only necessary tasks use cloud resources. For short or bursty jobs, Verlex can route workloads to serverless container environments such as Cloud Run, Fargate, ACI, and RunPod, providing rapid cold starts. For larger tasks, clusters with up to eight GPUs can be provisioned in a single line of code. Team features include a shared credit pool for organizations, with administrative controls for credit allocation.

The service is available through a private beta, with no credit card required to join. Pricing is transparent, with live price comparisons and per-second billing based on prepaid credits. Verlex positions itself as a cost-first routing solution for GPU compute, offering significant discounts through spot pricing and automating cloud infrastructure management for developers and teams.