Modal is cloud infrastructure for teams that develop, train and serve AI applications. It runs code on GPUs and CPUs in the cloud, and the site describes it as making cloud compute feel like local development. Functions run arbitrary code, scale with request volume, burst to thousands of GPU and CPU resources at peak, and drop back to zero when the work is done.
The platform covers three main areas. Inference serves open-weight and custom models through OpenAI-compatible endpoints. Training handles fine-tuning, reinforcement learning and evaluation on single-GPU or multi-node capacity. Sandboxes provide isolated environments for coding agents, reinforcement learning rollouts and untrusted code, and the site says users can create over a million Sandboxes in under a minute.
The site says Modal spans more than 20 clouds and routes each workload to suitable hardware in a suitable region. It also says users can get GPUs in seconds without quota requests or long lead times. For enterprise teams, the page lists workload isolation, role-based access control, audit logs, data residency controls, SOC 2 Type II compliance and HIPAA readiness.
Billing is by the second for the time code runs. The Starter plan is $0 plus compute per month with $30 of monthly free credits. The Team plan is $250 plus compute per month with $100 of monthly free credits. Enterprise pricing is custom. Early-stage startups and academic researchers can apply for free compute credits.
