Installation#

Install SkyPilot#

SkyPilot supports installation with uv or pip.

# Create a virtual environment with pip pre-installed (required for SkyPilot)
# SkyPilot requires 3.7 <= python <= 3.13.
uv venv --seed --python 3.10
source .venv/bin/activate  # Use WSL on Windows

uv pip install skypilot

# install dependencies for the clouds you want to use
uv pip install "skypilot[kubernetes,aws,gcp]"

Note

The --seed flag is required as it ensures pip is installed in the virtual environment. SkyPilot needs pip to build wheels for remote cluster setup.

# Install as a globally available tool with pip included
# SkyPilot requires 3.7 <= python <= 3.13.
uv tool install --with pip skypilot

# install dependencies for the clouds you want to use
uv tool install --with pip "skypilot[kubernetes,aws,gcp]"

Note

The --with pip flag is required when using uv tool install. Without it, SkyPilot will fail when building wheels for remote clusters.

# Recommended: use a new conda env to avoid package conflicts.
# SkyPilot requires 3.7 <= python <= 3.13.
conda create -y -n sky python=3.10
conda activate sky

pip install skypilot

# install dependencies for the clouds you want to use
pip install "skypilot[kubernetes,aws,gcp]"
Install SkyPilot from nightly build or source

SkyPilot provides nightly builds and source code for the latest features and for development.

Install from nightly build:

# Create a virtual environment with pip pre-installed (required for SkyPilot)
# SkyPilot requires 3.7 <= python <= 3.13.
uv venv --seed --python 3.10
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

uv pip install skypilot-nightly

# Build the dashboard (requires Node.js and npm)
npm --prefix sky/dashboard install
npm --prefix sky/dashboard run build

# Install as a globally available tool with pip included
# SkyPilot requires 3.7 <= python <= 3.13.
uv tool install --with pip skypilot-nightly

# Build the dashboard (requires Node.js and npm)
npm --prefix sky/dashboard install
npm --prefix sky/dashboard run build

# Recommended: use a new conda env to avoid package conflicts.
# SkyPilot requires 3.7 <= python <= 3.13.
conda create -y -n sky python=3.10
conda activate sky

pip install skypilot-nightly

# Build the dashboard (requires Node.js and npm)
npm --prefix sky/dashboard install
npm --prefix sky/dashboard run build

Install from source:

# Recommended: use a new conda env to avoid package conflicts.
# SkyPilot requires 3.7 <= python <= 3.13.
conda create -y -n sky python=3.10
conda activate sky

git clone https://github.com/skypilot-org/skypilot.git
cd skypilot

pip install -e .

Alternatively, we also provide a Docker image as a quick way to try out SkyPilot.

Run locally or connect to a remote API server#

SkyPilot can be run as a standalone application, or connect to a remote API server for multi-user collaboration.

To run SkyPilot locally:

Refer to the cloud setup section to download the necessary dependencies for the clouds you want to use.

Tip

You can install dependencies for multiple clouds at once with following commands:

# From stable release
uv pip install "skypilot[kubernetes,aws,gcp]"
# From nightly build
uv pip install "skypilot-nightly[kubernetes,aws,gcp]"

# From stable release
uv tool install --with pip "skypilot[kubernetes,aws,gcp]"
# From nightly build
uv tool install --with pip "skypilot-nightly[kubernetes,aws,gcp]"

# From stable release
pip install "skypilot[kubernetes,aws,gcp]"
# From nightly build
pip install "skypilot-nightly[kubernetes,aws,gcp]"
# From source
pip install -e ".[kubernetes,aws,gcp]"

Note

When using SkyPilot locally, run sky api stop after each upgrade or dependency installation to enable the new version. See Upgrading SkyPilot for more details.

To connect to a remote API server:

If your team has set up a SkyPilot remote API server, connect to it by running:

sky api login

There is no need to install any dependencies locally.

See Connecting to an API server for more details.

To deploy a remote API server:

See Deploying SkyPilot API Server for detailed instructions on how to deploy a remote API server.

Verify cloud access#

After installation, run sky check to verify that credentials are correctly set up:

sky check

This will produce a summary like:

Checking credentials to enable clouds for SkyPilot.
  AWS: enabled
  GCP: enabled
  Azure: enabled
  OCI: enabled
  Lambda: enabled
  Nebius: enabled
  RunPod: enabled
  Paperspace: enabled
  Fluidstack: enabled
  Cudo: enabled
  Shadeform: enabled
  IBM: enabled
  SCP: enabled
  Seeweb: enabled
  vSphere: enabled
  Cloudflare (for R2 object store): enabled
  Kubernetes: enabled
  Slurm: enabled

If any cloud’s credentials or dependencies are missing, sky check will output hints on how to resolve them. You can also refer to the cloud setup section below.

Tip

If your clouds show enabled — 🎉 🎉 Congratulations! 🎉 🎉 You can now head over to Quickstart to get started with SkyPilot.

Tip

To check credentials only for specific clouds, pass the clouds as arguments: sky check aws gcp

Tip

If you are having trouble setting up credentials, it may be because the API server started before they were configured. Try restarting the API server by running sky api stop and then sky api start.

Request quotas for first time users#

If your cloud account has not been used to launch instances before, the respective quotas are likely set to zero or a low limit. This is especially true for GPU instances.

Please follow Requesting Quota Increase to check quotas and request quota increases before proceeding.

Enable shell completion#

SkyPilot supports shell completion for Bash (Version 4.4 and up), Zsh and Fish. This is only available for click versions 8.0 and up (use pip install click==8.0.4 to install).

To enable shell completion after installing SkyPilot, you will need to modify your shell configuration. SkyPilot automates this process using the --install-shell-completion option, which you should call using the appropriate shell name or auto:

sky --install-shell-completion auto
# sky --install-shell-completion zsh
# sky --install-shell-completion bash
# sky --install-shell-completion fish

Shell completion may perform poorly on certain shells and machines. If you experience any issues after installation, you can use the --uninstall-shell-completion option to uninstall it, which you should similarly call using the appropriate shell name or auto:

sky --uninstall-shell-completion auto
# sky --uninstall-shell-completion zsh
# sky --uninstall-shell-completion bash
# sky --uninstall-shell-completion fish

Appendix: Cloud access for local SkyPilot#

SkyPilot can be run as a standalone application, or connect to a remote API server for multi-user collaboration.

When running SkyPilot locally, necessary dependencies and credentials need to be set up for the clouds you want to use. You can configure access to at least one cloud using the guides below.

To configure infra access for team deployment instead, see Optional: Configure cloud accounts.

SkyPilot supports most major cloud providers, Kubernetes, and Slurm.

Kubernetes#

Install the necessary dependencies for Kubernetes.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[kubernetes]"
# From nightly build
uv pip install "skypilot-nightly[kubernetes]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[kubernetes]"
# From nightly build
uv tool install --with pip "skypilot-nightly[kubernetes]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[kubernetes]"
# From nightly build
pip install "skypilot-nightly[kubernetes]"
# From source
pip install -e ".[kubernetes]"

SkyPilot can run workloads on on-prem or cloud-hosted Kubernetes clusters (e.g., EKS, GKE, Nebius Managed Kubernetes, Coreweave). The only requirement is a valid kubeconfig at ~/.kube/config.

# Place your kubeconfig at ~/.kube/config
mkdir -p ~/.kube
cp /path/to/kubeconfig ~/.kube/config

See SkyPilot on Kubernetes for more.

Tip

If you do not have access to a Kubernetes cluster, you can deploy a local Kubernetes cluster on your laptop with sky local up.

Slurm Early Access#

Note

Early Access: Slurm support is under active development. If you’re interested in trying it out, please fill out this form.

SkyPilot can run workloads on Slurm clusters. The only requirement is SSH access to a Slurm login node.

To configure Slurm support, create a ~/.slurm/config file with your Slurm cluster configuration and add the SSH credentials to connect to the Slurm login node.

# Create the Slurm config directory
mkdir -p ~/.slurm

# Add your Slurm cluster configuration
cat > ~/.slurm/config << EOF
Host mycluster
    HostName login.mycluster.myorg.com
    User myusername
    IdentityFile ~/.ssh/id_rsa
EOF

See SkyPilot on Slurm for more.

AWS#

Install the necessary dependencies for AWS.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[aws]"
# From nightly build
uv pip install "skypilot-nightly[aws]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[aws]"
# From nightly build
uv tool install --with pip "skypilot-nightly[aws]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[aws]"
# From nightly build
pip install "skypilot-nightly[aws]"
# From source
pip install -e ".[aws]"

To set up AWS credentials, log into the AWS console and create an access key for yourself. If you don’t see the “Security credentials” link shown in the AWS instructions, you may be using SSO; see Using AWS SSO.

Now configure your credentials.

# Configure your AWS credentials
aws configure

  • For AWS Access Key ID, copy the “Access key” value from console.

  • For the AWS Secret Access Key, copy the “Secret access key” value from console.

  • The Default region name [None]: and Default output format [None]: fields are optional and can be left blank to choose defaults.

To use AWS IAM Identity Center (AWS SSO), see here for instructions.

Optional: To create a new AWS user with minimal permissions for SkyPilot, see Dedicated SkyPilot IAM user.

GCP#

Install the necessary dependencies for GCP.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[gcp]"
# From nightly build
uv pip install "skypilot-nightly[gcp]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[gcp]"
# From nightly build
uv tool install --with pip "skypilot-nightly[gcp]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[gcp]"
# From nightly build
pip install "skypilot-nightly[gcp]"
# From source
pip install -e ".[gcp]"

# Install Google Cloud SDK via conda-forge
conda install -c conda-forge google-cloud-sdk

# Initialize gcloud
gcloud init

# Run this if you don't have a credentials file.
# This will generate ~/.config/gcloud/application_default_credentials.json.
gcloud auth application-default login

For MacOS with Silicon Chips:

curl -O https://dl.google.com/dl/cloudsdk/channels/rapid/downloads/google-cloud-cli-darwin-arm.tar.gz gcloud.tar.gz
tar -xf gcloud.tar.gz
./google-cloud-sdk/install.sh
# Update your path with the newly installed gcloud

If you are using other architecture or OS, follow the Google Cloud SDK installation instructions to download the appropriate package.

Be sure to complete the optional step that adds gcloud to your PATH. This step is required for SkyPilot to recognize that your gcloud installation is configured correctly.

Tip

If you are using multiple GCP projects, list all the projects by gcloud projects list and activate one by gcloud config set project <PROJECT_ID> (see GCP docs).

Common GCP installation errors

Here some commonly encountered errors and their fixes:

  • RemoveError: 'requests' is a dependency of conda and cannot be removed from conda's operating environment when running conda install -c conda-forge google-cloud-sdk — run conda update --force conda first and rerun the command.

  • Authorization Error (Error 400: invalid_request) with the url generated by gcloud auth login — install the latest version of the Google Cloud SDK (e.g., with conda install -c conda-forge google-cloud-sdk) on your local machine (which opened the browser) and rerun the command.

Optional: To create and use a long-lived service account on your local machine, see here.

Optional: To create a new GCP user with minimal permissions for SkyPilot, see GCP User Creation.

Azure#

Install the necessary dependencies for Azure.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
# Azure CLI has an issue with uv, and requires '--prerelease allow'.
uv pip install --prerelease allow azure-cli
uv pip install "skypilot[azure]"
# From nightly build
# Azure CLI has an issue with uv, and requires '--prerelease allow'.
uv pip install --prerelease allow azure-cli
uv pip install "skypilot-nightly[azure]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[azure]"
# From nightly build
uv tool install --with pip "skypilot-nightly[azure]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[azure]"
# From nightly build
pip install "skypilot-nightly[azure]"
# From source
pip install -e ".[azure]"

# Login
az login
# Set the subscription to use
az account set -s <subscription_id>

Hint: run az account subscription list to get a list of subscription IDs under your account.

CoreWeave#

CoreWeave integrates with SkyPilot through the Kubernetes integration. To set up:

  1. Install the necessary dependencies for CoreWeave.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[coreweave]"
# From nightly build
uv pip install "skypilot-nightly[coreweave]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[coreweave]"
# From nightly build
uv tool install --with pip "skypilot-nightly[coreweave]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[coreweave]"
# From nightly build
pip install "skypilot-nightly[coreweave]"
# From source
pip install -e ".[coreweave]"

  1. Launch a Coreweave CKS cluster from the CoreWeave console.

  2. Get your kubeconfig from the CoreWeave console and place it at ~/.kube/config.

Tip

CoreWeave also offers InfiniBand networking for high-performance distributed training. You can enable InfiniBand support by adding network_tier: best to your SkyPilot task configuration.

CoreWeave Object Storage (CAIOS)#

You can optionally set up CoreWeave Object Storage (CAIOS) as an S3-compatible object storage that can be used with SkyPilot for storing and accessing data in your workloads.

To get CAIOS Access Key ID and Secret Access Key:

  1. Log into your CoreWeave Cloud console.

  2. Navigate to Object Storage → Keys in the left sidebar.

  3. Generate a new key pair.

SkyPilot uses separate configuration files for CAIOS to avoid conflicts with your AWS credentials. Run the following command to configure your CAIOS access credentials:

AWS_SHARED_CREDENTIALS_FILE=~/.coreweave/cw.credentials aws configure --profile cw

When prompted, enter your CAIOS credentials:

AWS Access Key ID [None]: <your_access_key_id>
AWS Secret Access Key [None]: <your_secret_access_key>
Default region name [None]:
Default output format [None]: json

Next, configure the endpoint URL and addressing style for CoreWeave Object Storage. This tells AWS CLI how to connect to CoreWeave’s S3-compatible service:

# For external access (outside CoreWeave CKS clusters)
AWS_CONFIG_FILE=~/.coreweave/cw.config aws configure set endpoint_url https://cwobject.com --profile cw
AWS_CONFIG_FILE=~/.coreweave/cw.config aws configure set s3.addressing_style virtual --profile cw

Note

CAIOS offers two endpoints for different use cases. Choose the right endpoint:

  • External access (slow but accessible from anywhere): Use https://cwobject.com when launching SkyPilot clusters in non-CoreWeave CKS clusters. This endpoint is accessible from anywhere and uses secure HTTPS.

  • Internal access (fast but only accessible within CoreWeave’s network): Use http://cwlota.com only if you are launching SkyPilot clusters inside CoreWeave CKS clusters and do not need to upload local data to the bucket. The LOTA endpoint provides faster access within CoreWeave’s network but only supports HTTP and is not accessible externally. Refer to LOTA documentation for more details.

Nebius#

Nebius is the ultimate cloud for AI explorers. To configure Nebius access:

Install the necessary dependencies for Nebius.

Note

Nebius is only supported for Python >= 3.10

# Nebius requires 3.10 <= python <= 3.13.
# From stable release
uv pip install "skypilot[nebius]"
# From nightly build
uv pip install "skypilot-nightly[nebius]"

# Nebius requires 3.10 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[nebius]"
# From nightly build
uv tool install --with pip "skypilot-nightly[nebius]"

# Nebius requires 3.10 <= python <= 3.13.
# From stable release
pip install "skypilot[nebius]"
# From nightly build
pip install "skypilot-nightly[nebius]"
# From source
pip install -e ".[nebius]"

Install and configure Nebius CLI:

mkdir -p ~/.nebius
nebius iam get-access-token > ~/.nebius/NEBIUS_IAM_TOKEN.txt
nebius --format json iam whoami|jq -r '.user_profile.tenants[0].tenant_id' > ~/.nebius/NEBIUS_TENANT_ID.txt

Optional: You can specify specific project ID and fabric in ~/.sky/config.yaml, see Configuration project_id and fabric for Nebius.

Alternatively, you can also use a service account to access Nebius, see Using Service Account for Nebius.

To use Nebius Managed Kubernetes, see Kubernetes Installation. Retrieve the Kubernetes credential with:

nebius mk8s cluster get-credentials --id <cluster_id> --external --kubeconfig $HOME/.kube/config

Nebius also offers Object Storage, an S3-compatible object storage without any egress charges. SkyPilot can download/upload data to Nebius buckets and mount them as local filesystem on clusters launched by SkyPilot. To set up Nebius support, run:

# Install boto
pip install boto3
# Configure your Nebius Object Storage credentials
aws configure --profile nebius

In the prompt, enter your Nebius Access Key ID and Secret Access Key (see instructions to generate Nebius credentials). Select auto for the default region and json for the default output format.

aws configure set aws_access_key_id $NB_ACCESS_KEY_AWS_ID --profile nebius
aws configure set aws_secret_access_key $NB_SECRET_ACCESS_KEY --profile nebius
aws configure set region <REGION> --profile nebius
aws configure set endpoint_url <ENDPOINT>  --profile nebius

RunPod Community Maintained#

RunPod is a specialized AI cloud provider that offers low-cost GPUs. To configure RunPod access:

Install the necessary dependencies for RunPod

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[runpod]"
# From nightly build
uv pip install "skypilot-nightly[runpod]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[runpod]"
# From nightly build
uv tool install --with pip "skypilot-nightly[runpod]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[runpod]"
# From nightly build
pip install "skypilot-nightly[runpod]"
# From source
pip install -e ".[runpod]"

Go to the Settings page on your RunPod console and generate an API key. Then, run:

pip install "runpod>=1.6.1"
runpod config

OCI Community Maintained#

To access Oracle Cloud Infrastructure (OCI):

Install the necessary dependencies for OCI.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[oci]"
# From nightly build
uv pip install "skypilot-nightly[oci]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[oci]"
# From nightly build
uv tool install --with pip "skypilot-nightly[oci]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[oci]"
# From nightly build
pip install "skypilot-nightly[oci]"
# From source
pip install -e ".[oci]"

Setup the credentials by following this guide. After completing the steps in the guide, the ~/.oci folder should contain the following files:

~/.oci/config
~/.oci/oci_api_key.pem

The ~/.oci/config file should contain the following fields:

[DEFAULT]
user=ocid1.user.oc1..aaaaaaaa
fingerprint=aa:bb:cc:dd:ee:ff:gg:hh:ii:jj:kk:ll:mm:nn:oo:pp
tenancy=ocid1.tenancy.oc1..aaaaaaaa
region=us-sanjose-1
# Note that we should avoid using full home path for the key_file configuration, e.g. use ~/.oci instead of /home/username/.oci
key_file=~/.oci/oci_api_key.pem

By default, the provisioned nodes will be in the root compartment. To specify the compartment other than root, create/edit the file ~/.sky/config.yaml, put the compartment’s OCID there, as the following:

oci:
  region_configs:
    default:
      compartment_ocid: ocid1.compartment.oc1..aaaaaaaa......

Lambda Cloud Community Maintained#

Lambda Cloud is a cloud provider offering low-cost GPUs. To configure Lambda Cloud access:

Install the necessary dependencies for Lambda Cloud.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[lambda]"
# From nightly build
uv pip install "skypilot-nightly[lambda]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[lambda]"
# From nightly build
uv tool install --with pip "skypilot-nightly[lambda]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[lambda]"
# From nightly build
pip install "skypilot-nightly[lambda]"
# From source
pip install -e ".[lambda]"

Go to the API Keys page on your Lambda console to generate a key and then add it to ~/.lambda_cloud/lambda_keys:

mkdir -p ~/.lambda_cloud
echo "api_key = <your_api_key_here>" > ~/.lambda_cloud/lambda_keys

Together AI Community Maintained#

Together AI offers GPU instant clusters. Accessing them is similar to using Kubernetes:

  1. Install the necessary dependencies for Kubernetes.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[kubernetes]"
# From nightly build
uv pip install "skypilot-nightly[kubernetes]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[kubernetes]"
# From nightly build
uv tool install --with pip "skypilot-nightly[kubernetes]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[kubernetes]"
# From nightly build
pip install "skypilot-nightly[kubernetes]"
# From source
pip install -e ".[kubernetes]"

  1. Launch a Together Instant Cluster with cluster type selected as Kubernetes

  2. Get the Kubernetes config for the cluster

  3. Save the kubeconfig to a file, e.g., ./together.kubeconfig

  4. Copy the kubeconfig to your ~/.kube/config or merge the Kubernetes config with your existing kubeconfig file by running:

KUBECONFIG=./together-kubeconfig:~/.kube/config kubectl config view --flatten > /tmp/merged_kubeconfig && mv /tmp/merged_kubeconfig ~/.kube/config

Paperspace Community Maintained#

Paperspace is a cloud provider that provides access to GPU accelerated VMs. To configure Paperspace access:

Install the necessary dependencies for Paperspace.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[paperspace]"
# From nightly build
uv pip install "skypilot-nightly[paperspace]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[paperspace]"
# From nightly build
uv tool install --with pip "skypilot-nightly[paperspace]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[paperspace]"
# From nightly build
pip install "skypilot-nightly[paperspace]"
# From source
pip install -e ".[paperspace]"

Go to follow these instructions to generate an API key. Add the API key with:

mkdir -p ~/.paperspace
echo "{'api_key' : <your_api_key_here>}" > ~/.paperspace/config.json

Vast Community Maintained#

Vast is a cloud provider that offers low-cost GPUs. To configure Vast access:

Install the necessary dependencies for Vast.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[vast]"
# From nightly build
uv pip install "skypilot-nightly[vast]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[vast]"
# From nightly build
uv tool install --with pip "skypilot-nightly[vast]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[vast]"
# From nightly build
pip install "skypilot-nightly[vast]"
# From source
pip install -e ".[vast]"

Go to the Account page on your Vast console to get your API key. Then, run:

pip install "vastai-sdk>=0.1.12"
mkdir -p ~/.config/vastai
echo "<your_api_key_here>" > ~/.config/vastai/vast_api_key

Fluidstack Community Maintained#

Fluidstack is a cloud provider offering low-cost GPUs. To configure Fluidstack access:

Install the necessary dependencies for Fluidstack.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[fluidstack]"
# From nightly build
uv pip install "skypilot-nightly[fluidstack]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[fluidstack]"
# From nightly build
uv tool install --with pip "skypilot-nightly[fluidstack]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[fluidstack]"
# From nightly build
pip install "skypilot-nightly[fluidstack]"
# From source
pip install -e ".[fluidstack]"

Go to the Home page on your Fluidstack console to generate an API key and then add the API key to ~/.fluidstack/api_key :

mkdir -p ~/.fluidstack
echo "your_api_key_here" > ~/.fluidstack/api_key

Cudo Compute Community Maintained#

Cudo Compute provides low cost GPUs powered by green energy.

  1. Install the necessary dependencies for Cudo Compute.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[cudo]"
# From nightly build
uv pip install "skypilot-nightly[cudo]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[cudo]"
# From nightly build
uv tool install --with pip "skypilot-nightly[cudo]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[cudo]"
# From nightly build
pip install "skypilot-nightly[cudo]"
# From source
pip install -e ".[cudo]"

  1. Create a billing account.

  2. Create a project.

  3. Create an API Key.

  4. Download and install the cudoctl command line tool

  5. Run cudoctl init:

cudoctl init
  ✔ api key: my-api-key
  ✔ project: my-project
  ✔ billing account: my-billing-account
  ✔ context: default
  config file saved ~/.config/cudo/cudo.yml

pip install "cudo-compute>=0.1.10"

If you want to want to use SkyPilot with a different Cudo Compute account or project, run cudoctl init again.

Shadeform Community Maintained#

Shadeform is a cloud GPU marketplace that offers GPUs across a variety of vetted cloud providers. To configure Shadeform access:

Install the necessary dependencies for Shadeform.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[shadeform]"
# From nightly build
uv pip install "skypilot-nightly[shadeform]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[shadeform]"
# From nightly build
uv tool install --with pip "skypilot-nightly[shadeform]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[shadeform]"
# From nightly build
pip install "skypilot-nightly[shadeform]"
# From source
pip install -e ".[shadeform]"

Go to the API Key Management page within your Shadeform account to generate a key and then add it to ~/.shadeform/api_key:

mkdir -p ~/.shadeform
echo "<your_api_key_here>" > ~/.shadeform/api_key

IBM Community Maintained#

To access IBM’s VPC service:

Install the necessary dependencies for IBM.

Note

IBM is only supported for Python <= 3.11

# IBM requires 3.7 <= python <= 3.11.
# From stable release
uv pip install "skypilot[ibm]"
# From nightly build
uv pip install "skypilot-nightly[ibm]"

# IBM requires 3.7 <= python <= 3.11.
# From stable release
uv tool install --with pip "skypilot[ibm]"
# From nightly build
uv tool install --with pip "skypilot-nightly[ibm]"

# IBM requires 3.7 <= python <= 3.11.
# From stable release
pip install "skypilot[ibm]"
# From nightly build
pip install "skypilot-nightly[ibm]"
# From source
pip install -e ".[ibm]"

Store the following fields in ~/.ibm/credentials.yaml:

iam_api_key: <user_personal_api_key>
resource_group_id: <resource_group_user_is_a_member_of>

Note

Stock images aren’t currently providing ML tools out of the box. Create private images with the necessary tools (e.g. CUDA), by following the IBM segment in this documentation.

To access IBM’s Cloud Object Storage (COS), append the following fields to the credentials file:

access_key_id: <access_key_id>
secret_access_key: <secret_key_id>

To get access_key_id and secret_access_key use the IBM web console:

  1. Create/Select a COS instance from the web console.

  2. From “Service Credentials” tab, click “New Credential” and toggle “Include HMAC Credential”.

  3. Copy “secret_access_key” and “access_key_id” to file.

Finally, install rclone via: curl https://rclone.org/install.sh | sudo bash

Note

sky check does not reflect IBM COS’s enabled status. IBM: enabled only guarantees that IBM VM instances are enabled.

SCP (Samsung Cloud Platform) Community Maintained#

Samsung Cloud Platform, or SCP, provides cloud services optimized for enterprise customers. You can learn more about SCP here.

To configure SCP access:

Install the necessary dependencies for SCP.

Note

SCP is only supported for Python <= 3.11

# SCP requires 3.7 <= python <= 3.11.
# From stable release
uv pip install "skypilot[scp]"
# From nightly build
uv pip install "skypilot-nightly[scp]"

# SCP requires 3.7 <= python <= 3.11.
# From stable release
uv tool install --with pip "skypilot[scp]"
# From nightly build
uv tool install --with pip "skypilot-nightly[scp]"

# SCP requires 3.7 <= python <= 3.11.
# From stable release
pip install "skypilot[scp]"
# From nightly build
pip install "skypilot-nightly[scp]"
# From source
pip install -e ".[scp]"

You need access keys and the ID of the project your tasks will run. Go to the Access Key Management page on your SCP console to generate the access keys, and the Project Overview page for the project ID. Then, add them to ~/.scp/scp_credential by running:

# Create directory if required
mkdir -p ~/.scp
# Add the lines for "access_key", "secret_key", and "project_id" to scp_credential file
echo "access_key = <your_access_key>" >> ~/.scp/scp_credential
echo "secret_key = <your_secret_key>" >> ~/.scp/scp_credential
echo "project_id = <your_project_id>" >> ~/.scp/scp_credential

Note

Multi-node clusters are currently not supported on SCP.

VMware vSphere Community Maintained#

To configure VMware vSphere access:

Install the necessary dependencies for VMware vSphere.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[vsphere]"
# From nightly build
uv pip install "skypilot-nightly[vsphere]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[vsphere]"
# From nightly build
uv tool install --with pip "skypilot-nightly[vsphere]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[vsphere]"
# From nightly build
pip install "skypilot-nightly[vsphere]"
# From source
pip install -e ".[vsphere]"

Store the vSphere credentials in ~/.vsphere/credential.yaml:

mkdir -p ~/.vsphere
touch ~/.vsphere/credential.yaml

Here is an example of configuration within the credential file:

vcenters:
  - name: <your_vsphere_server_ip_01>
    username: <your_vsphere_user_name>
    password: <your_vsphere_user_passwd>
    skip_verification: true # If your vcenter have valid certificate then change to 'false' here
    # Clusters that can be used by SkyPilot:
    #   [] means all the clusters in the vSphere can be used by Skypilot
    # Instead, you can specify the clusters in a list:
    # clusters:
    #   - name: <your_vsphere_cluster_name1>
    #   - name: <your_vsphere_cluster_name2>
    clusters: []
  # If you are configuring only one vSphere instance, omit the following line.
  - name: <your_vsphere_server_ip_02>
    username: <your_vsphere_user_name>
    password: <your_vsphere_user_passwd>
    skip_verification: true
    clusters: []

After configuring the vSphere credentials, ensure that the necessary preparations for vSphere are completed. Please refer to this guide for more information: Cloud Preparation for vSphere

Cloudflare R2#

Cloudflare offers R2, an S3-compatible object storage without any egress charges. SkyPilot can download/upload data to R2 buckets and mount them as local filesystem on clusters launched by SkyPilot. To set up R2 support:

Install the necessary dependencies for Cloudflare R2.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[cloudflare]"
# From nightly build
uv pip install "skypilot-nightly[cloudflare]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[cloudflare]"
# From nightly build
uv tool install --with pip "skypilot-nightly[cloudflare]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[cloudflare]"
# From nightly build
pip install "skypilot-nightly[cloudflare]"
# From source
pip install -e ".[cloudflare]"

Run the following commands:

# Install boto
pip install boto3
# Configure your R2 credentials
AWS_SHARED_CREDENTIALS_FILE=~/.cloudflare/r2.credentials aws configure --profile r2

In the prompt, enter your R2 Access Key ID and Secret Access Key (see instructions to generate R2 credentials). Select auto for the default region and json for the default output format.

AWS Access Key ID [None]: <access_key_id>
AWS Secret Access Key [None]: <access_key_secret>
Default region name [None]: auto
Default output format [None]: json

Next, get your Account ID from your R2 dashboard and store it in ~/.cloudflare/accountid with:

mkdir -p ~/.cloudflare
echo <YOUR_ACCOUNT_ID_HERE> > ~/.cloudflare/accountid

Note

Support for R2 is in beta. Please report and issues on Github or reach out to us on Slack.

Prime Intellect Community Maintained#

Prime Intellect makes it easy to find global compute resources and train state-of-the-art models through distributed training across clusters. To configure Prime Intellect access:

Install the necessary dependencies for Prime Intellect.

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv pip install "skypilot[primeintellect]"
# From nightly build
uv pip install "skypilot-nightly[primeintellect]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[primeintellect]"
# From nightly build
uv tool install --with pip "skypilot-nightly[primeintellect]"

# SkyPilot requires 3.7 <= python <= 3.13.
# From stable release
pip install "skypilot[primeintellect]"
# From nightly build
pip install "skypilot-nightly[primeintellect]"
# From source
pip install -e ".[primeintellect]"

Install and configure Prime Intellect CLI:

mkdir -p ~/.prime
prime login
# optional: set team id
prime config set-team-id <team_id>

Seeweb Community Maintained#

Seeweb is your European GPU Cloud Provider. To access Seeweb:

  1. Install the necessary dependencies for Seeweb.

Note

Seeweb is only supported for Python >= 3.10

# Seeweb requires 3.10 <= python <= 3.13.
# From stable release
uv pip install "skypilot[seeweb]"
# From nightly build
uv pip install "skypilot-nightly[seeweb]"

# Seeweb requires 3.10 <= python <= 3.13.
# From stable release
uv tool install --with pip "skypilot[seeweb]"
# From nightly build
uv tool install --with pip "skypilot-nightly[seeweb]"

# Seeweb requires 3.10 <= python <= 3.13.
# From stable release
pip install "skypilot[seeweb]"
# From nightly build
pip install "skypilot-nightly[seeweb]"
# From source
pip install -e ".[seeweb]"

  1. Log into your Seeweb dashboard :.

  2. Navigate to Compute → API Token in the control panel, and create New TOKEN.

  3. Create the file ~/.seeweb_cloud/seeweb_keys with the following contents:

[DEFAULT]
api_key = <your-api-token>

Appendix: Using SkyPilot in Docker#

As a quick alternative to installing SkyPilot on your laptop, we also provide a Docker image with SkyPilot main branch automatically cloned. You can simply run:

# NOTE: '--platform linux/amd64' is needed for Apple silicon Macs
docker run --platform linux/amd64 \
  -td --rm --name sky \
  -v "$HOME/.sky:/root/.sky:rw" \
  -v "$HOME/.aws:/root/.aws:rw" \
  -v "$HOME/.config/gcloud:/root/.config/gcloud:rw" \
  berkeleyskypilot/skypilot

docker exec -it sky /bin/bash

If your cloud CLIs are already set up, your credentials (AWS and GCP) will be mounted to the container and you can proceed to Quickstart. Otherwise, you can follow the instructions in Cloud account setup inside the container to set up your cloud accounts.

Once you are done with experimenting with SkyPilot, remember to delete any clusters and storage resources you may have created using the following commands:

# Run inside the container:
sky down -a -y
sky storage delete -a -y

Finally, you can stop the container with:

docker stop sky

See more details about the dev container image berkeleyskypilot/skypilot-nightly here.