Compute API
Use the compute API to list, launch, stop, and terminate multi-provider GPU instances.
GET /v1/compute
List all instances visible through your connected provider keys.
curl https://api-staging.pulserun.dev/v1/compute \
-H "Authorization: Bearer YOUR_API_KEY"POST /v1/compute/launch
Launch a new instance on a specific provider.
curl -X POST https://api-staging.pulserun.dev/v1/compute/launch \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"provider": "runpod",
"gpu_type": "a100_80gb",
"gpu_count": 1,
"docker_image": "runpod/pytorch:2.1.0-py3.10-cuda12.1.0-devel-ubuntu22.04",
"spot": false,
"name": "training-a100"
}'Request fields
| Field | Type | Required | Notes |
|---|---|---|---|
provider | string | Yes | runpod, vast, or lambda |
gpu_type | string | Yes | Normalized GPU slug |
gpu_count | integer | No | Default 1 |
docker_image | string | No | Image to launch |
spot | boolean | No | Spot when supported |
name | string | No | Instance name |
volume_gb | integer | No | RunPod volume size |
disk_gb | integer | No | Disk size |
region | string | No | Required for Lambda |
ssh_key_name | string | No | Lambda SSH key name |
onstart_cmd | string | No | Vast startup command |
POST /v1/compute/{provider_instance_id}/stop
Stop an instance.
curl -X POST https://api-staging.pulserun.dev/v1/compute/<provider_instance_id>/stop \
-H "Authorization: Bearer YOUR_API_KEY"Stop is currently supported for RunPod instances only.
POST /v1/compute/{provider_instance_id}/terminate
Terminate an instance.
curl -X POST https://api-staging.pulserun.dev/v1/compute/<provider_instance_id>/terminate \
-H "Authorization: Bearer YOUR_API_KEY"