Tool profile · Publisher documentation
Modal is a serverless cloud platform that lets teams deploy, train and serve AI applications on GPU infrastructure billed by the second. According to its supplier's public documentation, it covers model inference, isolated sandboxes, GPU training, serverless autoscaling and distributed storage. LE LABO IA has not tested the product; this profile relies on the public documentation of its supplier. Pricing is documented with a Starter plan at 0 dollars plus compute per month and a Team plan at 250 dollars plus compute per month, while Enterprise pricing is custom. GPU rates range from 0.000164 dollars per second for an Nvidia T4 to 0.001972 dollars per second for an Nvidia B300, and CPU, memory, volumes and network egress are also billed. Territorial availability, quotas and terms are not confirmed, so check them with the supplier before committing. This profile is meant for readers choosing a tool, and it leaves unknown information unknown.
Tool profile · Publisher documentation
The product points below come from publisher documentation. No common hands-on test is claimed. Documentary profile, sources checked October 7, 2026. LE LABO IA has not tested this product.
Deploy AI applications on a serverless cloud infrastructure that bills GPU usage by the second, as described in the supplier's documentation. Train models on GPU capacity that scales automatically instead of managing dedicated machines. Serve model inference through isolated sandboxes suited to running workloads with separated execution environments. Rely on distributed storage for data and artifacts used across training and serving jobs.
Starter: 0 dollars plus compute per month, with 30 dollars per month of credits, 3 seats, 100 containers and 10 concurrent GPUs. Team: 250 dollars plus compute per month, with 100 dollars per month of credits, unlimited seats, 5000 containers and 50 concurrent GPUs. Enterprise: custom pricing. GPU usage is billed by the second: Nvidia B300 0.001972 dollars/sec, B200 0.001736, H200 SXM 0.001261, H100 SXM5 0.001097, RTX PRO 6000 0.000842, A100 80 GB 0.000694, A100 40 GB 0.000583, L40S 0.000542, A10 0.000306, L4 0.000222 and T4 0.000164. CPU is 0.0000131 dollars per core per second, memory 0.00000222 dollars per GiB per second, volumes 0.09 dollars per GiB per month and network egress 0.04 dollars per GiB. These amounts are not confirmed and should be checked with the supplier.
LE LABO IA has not tested the product, so the profile rests only on the public documentation of its supplier. Quotas, territorial availability and terms of use are not confirmed and should be checked with the supplier. Prices come from the supplier's pricing documentation, where GPU and CPU rates are shown per second with an hourly view also available.
Before committing, you can open a free Starter account and deploy a small workload to see how the documented flow behaves in practice. You can then check the supplier's site for the GPU rates that apply to your region and confirm whether the quotas, seats and containers listed for each plan match your needs. It is also worth verifying the terms of use and the availability of the service in your country directly with the supplier.
