Modal

A cloud infrastructure platform for running Python code on serverless GPUs and CPUs, designed specifically for machine learning inference, model training, and AI data processing workloads. Developers write regular Python functions and decorate them with Modal-specific decorators; Modal then handles container building, dependency installation, GPU provisioning, autoscaling from zero to hundreds of parallel workers, and job scheduling without any infrastructure management. This eliminates the need to configure Kubernetes, manage cloud VMs, or write Dockerfiles for AI workloads. Modal is especially popular for deploying AI models as production APIs, running Stable Diffusion, Flux, Whisper, and custom fine-tuned models on demand, processing large datasets in parallel, and running scheduled batch jobs. GPU pricing starts at $0.00030 per GPU-second, approximately $1.08 per hour for an A10G, and scales to A100 and H100 instances. A free tier provides $30 in credits per month, which covers significant experimentation. Popular with ML engineers, AI startup teams, and developers who want production-grade GPU infrastructure without DevOps overhead.

Browse tasks in this category · Category overview