vectorlay 0.1.0

The official Python SDK for the VectorLay API

VectorLay Python SDK

The official Python SDK for the VectorLay API.

Installation

pip install vectorlay

Quick Start

import vectorlay

client = vectorlay.Client(api_key="vl_your_api_key")

# List available GPUs
gpus = client.gpus.list()

# Deploy a cluster
cluster = client.clusters.create(
    name="my-inference",
    gpu_type="h100",
    replicas=2,
    container="vllm/vllm-openai:latest",
    env={"MODEL": "meta-llama/Llama-3-8B"},
    container_port=8000,
)

# Wait for it to be ready
cluster = client.clusters.wait_for_status(cluster.id)
print(f"Cluster running at: {cluster.endpoint_url}")

# Scale up
client.clusters.scale(cluster.id, replicas=4)

# Create a VM
vm = client.vms.create(
    name="dev-box",
    ssh_key_ids=["your-ssh-key-id"],
    gpu_type="rtx-4090",
    gpu_count=1,
    disk_size_gb=100,
)

Async Usage

import asyncio
import vectorlay

async def main():
    async with vectorlay.AsyncClient(api_key="vl_your_api_key") as client:
        clusters = await client.clusters.list()
        print(clusters)

asyncio.run(main())

Configuration

The client accepts these parameters:

Parameter	Env Variable	Default
api_key	VECTORLAY_API_KEY	Required
base_url	VECTORLAY_BASE_URL	https://api.vectorlay.com
timeout	—	60.0 seconds
max_retries	—	2

License

MIT