Overview
Nebula Block documentation — serverless inference, GPU cloud, and S3-compatible object storage, with Canadian-hosted options.
Welcome to Nebula Block — a Montreal-based AI cloud offering serverless inference, on-demand GPU compute, and S3-compatible object storage, with Canadian-hosted options for workloads that have data-residency requirements.
These docs cover everything from your first API call to the full REST reference.
Start here
Make your first API call in a few minutes
Understand what Nebula Block offers
See which models you can call
Know your rate limits
Look up an endpoint
Core products
Serverless Inference
Call state-of-the-art models — text, vision, image, video, embedding, and reranking — through a single OpenAI-compatible endpoint at https://inference.nebulablock.com/v1. No servers to provision, no models to load; point any OpenAI SDK at it and go.
→ Serverless Inference · Model Catalog
GPU Cloud
Rent on-demand GPU and CPU instances by the hour, from single GPUs to multi-GPU nodes, deployed in minutes with your own SSH keys and firewall rules.
Object Storage
S3-compatible storage for datasets, checkpoints, and model outputs. Works with s3cmd, boto3, and the AWS SDKs you already use.
Build with the API
Nebula Block exposes two APIs:
Inference API (OpenAI-compatible)
https://inference.nebulablock.com/v1
Chat completions, vision, images, video, embeddings, reranking
Platform API
https://api.nebulablock.com/api/v1
Instances, SSH keys, API keys, object storage, billing, teams
→ API Reference · Authentication
Manage your account
API Keys — create and rotate keys
Tiers and Rate Limits — what your tier unlocks
Teams — shared billing, roles, and team API keys
Referral Program — earn commission on referred spend
Get help
Last updated