Crusoe for AI developers

The AI builder’s cloud

Everything you need to train, fine-tune, and serve your models on Crusoe Cloud.

Choose your path

Crusoe Intelligence Foundry

Managed inference for AI applications. Run open models, bring your own, or fine-tune — no infrastructure overhead.

Crusoe Cloud

GPU compute and orchestration for large-scale training and inference workloads. Full control of your stack.

Connect with us
at events

Yotta 2026
Las Vegas, NV
Inside the cluster: How AI teams actually orchestrate at scale
AI Infra Summit
Santa Clara, CA, USA

Your toolkit

Documentation

Quickstart guides and API reference for Compute, Storage, Networking, Orchestration, and Managed Inference.

Learn more

How-to guides

Step-by-step walkthroughs for common tasks — from spinning up your first instance to configuring managed inference, storage, networking, and more.

Learn more

Support

Direct technical support from our team, whenever you need it.

Learn more

Demos

Real examples from our GitHub and Hugging Face Spaces. Run the demos, fork the code.

01

Model Arena

02

FakePod

03

Snake Battle AI

04

TorchTitan Pretraining on CMK

05

PyTorch Distributed Training on SLURM

06

Ray Inference on CMK

Build with us

Where Crusoe engineers and AI builders hang out. See what's shipping and get the first look at new models, hackathons, and previews.

From our engineering team

Isometric illustration of dense server racks representing NVIDIA HGX B300 systems on Crusoe Cloud in Iceland
September 21, 2026
Crusoe earns NVIDIA Exemplar Cloud validation on Blackwell Ultra
Isometric illustration of server racks representing Crusoe's AMD MI355X MLPerf Inference cluster
September 16, 2026
Serving 5.75 million tokens per second: Crusoe's MLPerf Inference v6.1 results on AMD MI355X
Isometric illustration of server racks representing NVIDIA GB200 NVL72 on Crusoe Cloud for MLPerf Inference v6.1
September 16, 2026
Crusoe's MLPerf Inference v6.1 results on NVIDIA GB200 NVL72
[Crusoe] infrastructure played a massive role in this [AI DevSummit Hackathon] win. I used NVIDIA Nemotron via Crusoe to reliably generate personalized skincare routines based on complex environmental and skin analysis data.
Perbangga Duha
Hackathon winner
~250 Crusoe calls, 0 user-visible failures…it was rock solid.
Alexander Sorrell
Hackathon participant
I built Leavitt, an on-call AI SRE agent for diagnosing production incidents. Running Nous Research Hermes + NVIDIA Nemotron on Crusoe Cloud, it held its line under chaos.
Alexander Sorrell
Hackathon winner

Frequently
asked questions

Ready to start building?