Run Models

Use run to call a model and wait for the result. Parameters are passed with repeatable --param key=value flags. Complex values should be passed as JSON strings; nested fields can also use dot notation.

Parameter format

The run command accepts repeatable --param flags. Arrays and objects should be passed as JSON strings.

seacloud run <model_id> --param key=value
seacloud run <model_id> --param object.field=value
seacloud run <model_id> --param array='[{"type":"text","text":"hello"}]'

Image generation

Image generation examples use gpt-image-2. The model can be called through run and can output result URLs.

seacloud run gpt-image-2 --param prompt="a blue cat" --output url

Video generation

Video generation examples use seedance_2_0. The model can be called through run and can output result URLs.

seacloud run seedance_2_0 --param content='[{"type":"text","text":"a cat running"}]' --param duration=5 --output url

Output formats

Option	Description
--output url	Prints only generated result URLs
--output json	Prints the full structured response
--timeout <seconds>	Sets the maximum wait time for results. The default is 600 seconds