dstack

A unified orchestration layer for heterogeneous AI compute. It standardizes how to manage compute and run training and inference on GPU clouds, Kubernetes, VMs,

What is dstack?

A unified orchestration layer for heterogeneous AI compute. It standardizes how to manage compute and run training and inference on GPU clouds, Kubernetes, VMs, or bare-metal clusters.

Overview of dstack

What it is

dstack is a unified control plane for GPU provisioning and orchestration that works with any GPU cloud, Kubernetes, or on-prem clusters. It standardizes how to manage compute and run training and inference on GPU clouds, Kubernetes, VMs, or bare-metal clusters, and is compatible with any hardware, open-source tools, and frameworks. NVIDIA, AMD, Google TPU, and Tenstorrent accelerators are supported out of the box. The server can be installed on Linux, macOS, and Windows (via WSL 2), and requires Git and OpenSSH.

Key features

  • Supports NVIDIA, AMD, Google TPU, and Tenstorrent accelerators out of the box.
  • Configuration types include fleets, dev environments, tasks, services, presets, and volumes, defined as YAML files within your repo.
  • Works across GPU clouds, Kubernetes, and on-prem clusters; with on-prem servers, backend configuration is not required and SSH fleets are created once the server is up.
  • Automatically manages provisioning, job queuing, auto-scaling, networking, volumes, run failures, out-of-capacity errors, and port-forwarding across clouds and on-prem clusters.
  • Configurations can be applied via the dstack apply CLI command, a programmatic API, or AI agent skills.
  • Agent skills let AI agents such as Claude, Codex, and Cursor create and manage fleets and submit workloads.

Use cases

  • Fleets: provision and manage clusters across clouds, Kubernetes, and on-prem.
  • Dev environments: launch environments accessed by agents or from an IDE.
  • Tasks: run training, batch, or other jobs across a single node or clusters.
  • Services: deploy model inference as secure and scalable endpoints.
  • Volumes: manage instance and network volumes for persisting data.

Getting started

shell$ uv tool install \”dstack[all]\” -U$ dstack serverOnce the server is running, install the CLI if it was not installed with the server, and point it at the server:shell$ uv tool install dstack -U$ dstack project add \ --name main \ --url http://127.0.0.1:3000 \ --token bbae0f28-d3dd-4820-bf61-8f4bb40815daAgent skills can be installed with npx skills add dstackai/dstack.

License and pricing

Mozilla Public License 2.0. The project is open source; no pricing is stated in the README or metadata.

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/dstackaidstack.svg)](https://yunodb.com/tools/dstackaidstack?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/dstackaidstack?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/dstackaidstack.svg" alt="Featured on YunoDB" height="20"></a>