Modal
Modal is a serverless cloud platform built for AI and data teams who want to run Python code on CPUs and GPUs without managing infrastructure. You define logic, hardware, and container images in code through the Modal SDK, then deploy inference, training, batch jobs, sandboxes, and notebooks from a single workflow.
The platform is designed around heavy AI workloads: sub-second cold starts, instant autoscaling from zero to thousands of GPUs, and per-second billing so you only pay while code is running. Modal built its own file system, container runtime, scheduler, and image builder to deliver that experience.
Teams use Modal for generative AI inference, LLM fine-tuning, agent sandboxes, computational biology, and media processing. Customers on the site include Runway, Suno, Decagon, Quora, Lovable, and Substack.
Define cloud logic, hardware, and container images in Python without Dockerfiles or YAML
Sub-second cold starts with autoscaling from zero to 1000+ GPUs across regions
First-class workloads for inference, training, batch processing, sandboxes, and notebooks
Per-second billing with no idle charges and $30 monthly free compute on the Starter plan
Built-in logging, metrics, and observability across functions, sandboxes, and containers
Usage-based per-second billing avoids paying for idle GPU capacity.
Python-native SDK keeps infrastructure, hardware, and deployment in one codebase.
Strong customer proof points across inference, training, and agent workloads.
Primary SDK is Python-focused, which may not suit teams standardized on other languages.
Starter plan caps concurrency and scheduled functions compared with paid tiers.
How does Modal pricing work?
Modal charges per second for actual compute time on CPUs, GPUs, and memory. Workspace plans add included credits: Starter includes $30 per month, Team includes $100 per month, and Enterprise is custom.
Is there a free tier on Modal?
Yes. The Starter plan costs $0 plus compute usage and includes $30 per month in free credits, up to 3 workspace seats, and 100 containers with 10 GPU concurrency.
What GPUs does Modal support?
Modal offers NVIDIA GPUs including B300, B200, H200, H100, A100, L40S, A10, L4, and T4. Pricing is listed per second on the pricing page.
What programming language does Modal use?
Modal is built around a Python SDK. You stay in Python to define functions, hardware, storage, scheduled jobs, and HTTPS endpoints.
Does Modal offer startup or academic credits?
Yes. Modal offers credit grants for early-stage startups and up to $10,000 in credits for graduate students, labs, and researchers through separate application pages.

