Skip to main content

Machine Types

With fal-serverless, you can choose different sizes of machines to run each of your models. You are charged only for what you use.

To use different types of machines:

$ vi models/

def model(dbt, fal):
dbt.config(fal_environment="ml", fal_machine="GPU") # Add this line

df: pd.DataFrame = dbt.ref("orders_daily")

The following options are available for the fal_machine argument:

XS0.25 CPU cores, 256 mb RAM (default)
S0.50 CPU cores, 1 gb RAM
M2 CPU cores, 8 gb RAM
L4 CPU cores, 32 gb RAM
XL8 CPU cores, 128 gb RAM
GPU8 CPU cores, 64 gb RAM, 1 GPU core (a100, 40 gb)