modal - Run Python workloads on Modal GPUs
Deploy and run Python workloads on Modal serverless GPUs, including inference APIs, batch jobs, scheduled tasks, volumes, secrets, and cost estimation.
Tags
Updated: 2026-09-29Capabilities
What this skill does
- Deploy Modal applications
- Run GPU functions
- Serve inference endpoints
- Run parallel batch jobs
- Schedule recurring jobs
- Manage persistent volumes
- Inject application secrets
- Build execution images
- Select GPU types
- Estimate cloud costs
Inputs
- Python application code
- GPU workload parameters
- Model files
- Input data
- Modal volume names
- Secret names
- GPU type
- Job schedule
Outputs
- Deployed Modal applications
- HTTPS inference endpoints
- Inference results
- Batch processing results
- Scheduled job executions
- Persistent volume data
- Application status
- Cloud usage estimates
Requirements
- Python environment
- Modal Python package
- Modal authentication
- Modal cloud access