Modal Serverless Compute Platform

Offering overview

Code is deployed as decorated Python functions; the platform builds containers, provisions the requested GPU or CPU and scales from zero to thousands of workers in seconds. Supported accelerators include NVIDIA T4, L4, A10, L40S, A100 (40 and 80 GB), H100, H200, RTX PRO 6000, B200 and B300. Beyond inference endpoints, the runtime offers sandboxes for executing untrusted agent-generated code, notebooks, scheduled jobs, batch processing and persistent volumes. Workloads can be pinned to broad regions (us, eu, ap) or narrow ones such as Virginia, Oregon, Dublin, Mumbai, Japan, Australia, the UK, Canada and Mexico, at a 1.15 to 1.75x price multiplier. ML engineers and AI startups building agents, inference services and data pipelines are the main users; the team works out of New York, Stockholm and San Francisco, with an additional London office. Compute is metered per second: an NVIDIA T4 costs $0.000164 per second (about $0.59 per hour), an H100 about $3.95 per hour and a B200 about $6.25 per hour, plus CPU and memory charges. The Starter plan has no platform fee and includes $30 of monthly credits, while Team adds $250 per month for higher limits.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

Not stated

AI frameworks

Not stated

Inference endpoints

Unknown

Fine-tuning

Unknown

Vector database

Unknown

Model catalog

Unknown

Monitoring

Not stated

Evaluation

Not stated

Security & compliance

Geography

Provider headquarters are separate from offering infrastructure and availability.

Available plans

Pricing can change. Promotional prices are shown separately from standard prices.