What is Modal
Modal is a serverless compute platform for running AI, data and Python workloads with usage-based CPU, memory, GPU and storage billing.
Plans
As of July 31, 2026, Starter costs US$0 plus compute and includes US$30 of monthly compute credit, three seats, 100 containers and GPU concurrency of 10. Team costs US$250 per month plus compute, includes US$100 of monthly credit, unlimited seats, 1,000 containers, GPU concurrency of 50, unlimited scheduled functions, custom domains, static proxying and rollbacks. Enterprise uses custom pricing and adds volume discounts, higher concurrency, ML engineering and Slack support.
Compute pricing
Published GPU rates per second include US$0.000164 for T4, US$0.000222 for L4, US$0.000306 for A10, US$0.000694 for A100 80GB, US$0.001097 for H100, US$0.001261 for H200 and US$0.001736 for B200. CPU costs US$0.0000131 per core-second and memory costs US$0.00000222 per GiB-second. Modal states that compute is billed only while code is running, without idle-time charges.
Volume storage costs US$0.09 per GiB-month and includes one TiB of data transfer per month. Sandbox and notebook resources use separately published CPU and memory rates.
Official source: Modal pricing.