Using the CLI

Create Inference Service

Our platform always selects the least expensive instance type and optimizes for your provided requirements. You can use any container image you want: private and public. If the image is private, please remember to first add your registry via the CLI. As of now, we only support workloads whith CUDA Version 12.3.

Create an inference service deploy an nginx container running on two CPU cores and serving requests on port 80:

perian inference create --image nginx --cores 2 --container-port 80

Create an inference service with more specific hardware requirements:

perian inference create --image nginx --cores 8 --memory 45 --accelerator-type V100 --container-port 80

Create an inference service from a model template:

perian inference create --model-template-id [model-template-id]

Get Inference Service Details

Display all created Inference services for your account:

perian inference get --all

Display information for a specific inference service:

perian inference get [inference-service-id]

Replace [inference-service-id] with a UUID of an inference service you have previously created or use the flag --last.

Cancel Inference Service

Cancel an inference service you have started:

perian inference cancel [inference-service-id]

Replace [inference-service-id] with a UUID of an inference service you have previously created.

This will cancel the inference service immediately.

Create Jobs

Create a job on an instance type with 2 CPU cores:

perian job create --image hello-world --cores 2

Create a job with more specific hardware requirements:

perian job create --image hello-world --cores 8 --memory 45 --accelerator-type V100

Get Job Details

Display all created jobs for your account:

perian job get --all

Display information for a specific job:

perian job get [job-id]

Replace [job-id] with a UUID of a job you have previously created or use the flag --last.

Cancel Jobs

Cancel a job you have started:

perian job cancel [job-id]

Replace [job-id] with a UUID of a job you have previously created.

This will cancel the job immediately.

Browse Model Templates

Browse through all available predefined model templates.

perian template get
Search for a model template by name: 
perian template get --name [model-template-name]

Browse Instance Types

Browse through all available instance types.

Search for specific number of cores and gigabytes of ram:

perian instance-type get --cores 4 --memory 16
Search for number of accelerators (GPUs): 
perian instance-type get --accelerators 2
Search for specific accelerator type: 
perian instance-type get --accelerator-type A100

perian instance-type get --accelerators 2 --accelerator-type A100

Browse Accelerator Types

Get a list of all available accelerator types:

perian accelerator-type get --all

Search for specific amount of accelerator memory:

perian accelerator-type get --memory 32

Get Billing Information

You can always check your total consumption and bill for the latest month:

perian billing get

You can also specify a specific timeframe to get the bill for:

perian billing get --start-time 2024-05-13 --start-time 2024-05-14

Getting Help

If you encounter any issues or have questions, please contact our support team at [email protected]. Feel free to ask questions in our Slack community and exchange thoughts and feedback with us and other users.