launch-logo

Launch

Ship and Track Your Models in Production

Scale Launch enables you to instantly ship your models to production with peace of mind.


  • Seamlessly provision and scale CPU and GPU compute
  • Manage the entire lifecycle of models—shadow deployments, A/B testing, and rollback
  • Access ML-aware monitoring with real-time metrics on uptime, latency, performance
1client = launch.LaunchClient()
2model_endpoint = client.create_endpoint (
3  model_bundle=my_pytorch_model,
4  **service_params
5)
6
7predictions = client.predict (
8  model_endpoint, 
9  image
10)
Run code