Active
Modal
Serverless compute for AI models, batch jobs, and isolated agent execution.
- Python-defined functions and compute workloads
- GPU options for inference and training
- Sandboxes for running generated or untrusted code
Modal runs functions, containers, and compute-heavy workflows on managed infrastructure. Applications declare the image, processor, memory, secrets, and storage they need. Sandboxes provide a separate execution environment for workloads such as agent-generated code.
How pricing works
Compute is priced by resource and running time, including GPU, CPU, and memory usage. Volumes are billed by stored capacity. Workspace plans can add a monthly fee on top of compute, so compare the whole workload and plan.
Sources
Checked .