Docs
Deploy AI

APIPod

APIPod packages a Python function into a production-ready service. CUDA detection, job queue, deploy artifacts. Start with the four-step quickstart, then dig into the command and deployment references.

Model as a Service

Your Service. Your Model. Any Cloud.

We are on a mission to make AI models and services accessible to everyone, everywhere without the headache. Deploy privately or publish your API.

Your Code

Socaity MaaS

RunPod

Scaleway (planned)

Azure (planned)

Deploy Privately

Your models, your data, your infrastructure. Full control with private deployments.

Smart Routing

Choose your cloud provider and region. Deploy to RunPod, Scaleway, or Azure.

Active-Only Billing

Pay for inference time only. Zero cost for idle, boot, or failed attempts.

No Lock-In

Standard Docker packaging. Your code stays portable across any provider.

Start here

Getting Started

Install APIPod and write your first service in four steps: install, write service.py, serve locally, test the endpoint.

Build

Package your service into a production Docker image with socaity build. CUDA detection, layer caching, custom Dockerfile support.

Endpoints

@app.endpoint reference: decorator parameters, media file types, progress reporting, queue behaviour.

Deploy Serverless

Deploy as a serverless GPU endpoint that scales to zero when idle and only bills while actively running.

Deploy Dedicated

Deploy on always-on GPU hardware for predictable latency and steady, high-volume workloads.

Guide

Host an Existing Model

Wrap an open Hugging Face or Replicate model in a FastTaskAPI service and self-host it on Socaity GPUs: build, push, deploy, and size the GPU.

Providers

GPU cloud providers supported by APIPod. RunPod EU is live today; Scaleway and Azure are coming soon.

Quickstart

Getting Started