Developers

Start here

Sign up, get a key, run one command. The full reference lives in the docs site; this is the shortest path from zero to a running GPU.

Quickstart

Zero to GPU

STEP 01

Create an account

A company email and a payment method. Verification is automatic for most domains; regulated sectors get a short KYC step instead.

STEP 02

Issue an API key

Scoped to a project and a role. Keys can be rotated, expired and audited from the console or the API.

STEP 03

Launch a pod

One command with a GPU type, a count and an image. SSH details and a Jupyter URL come back in the response.

rest: create a pod

$ curl -X POST https://api.mslproducts.com/v1/pods \
  -H "Authorization: Bearer $MSL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "gpu": "l40s",
    "count": 1,
    "image": "msl/vllm:0.6",
    "region": "pnq1",
    "idle_timeout_s": 900
  }

{ "id": "pod-3c9x1a", "state": "starting" }

terraform: a reserved cluster

resource "msl_cluster" "training" {
  name       = "llama-ft"
  region     = "bom1"
  gpu_type   = "h100-sxm"
  node_count = 8
  fabric     = "infiniband-400g"
  scheduler  = "slurm"

  volume {
    tier = "shared-fs"
    size = "120Ti"
  }
}

Reference

The rest of it

↗API

Start here

Zero to GPU

Create an account

Issue an API key

Launch a pod

The rest of it

REST reference

Python and Go

Terraform provider

Base containers

Migration guides

Status and incidents