#2929th of 168 in ML frameworks and libraries
skypilot
Run and scale AI jobs across Kubernetes, Slurm and 20+ clouds
- Stars
- 10.8k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Oct 2026
- Language
- Python
9,789 on 21 Apr 202610,821 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
SkyPilot takes a YAML or Python task spec (resources, workdir, setup, run) and launches it on the cheapest available infra among your Kubernetes clusters, Slurm clusters or cloud accounts. It provisions GPUs, TPUs or CPUs with auto-failover, syncs code, runs setup and streams logs. It also handles job queueing, autostop of idle clusters, and a shared API server for teams.
Who it is for: AI and infra teams running GPU jobs across clouds, Kubernetes or Slurm
Strengths
- One task spec runs on Kubernetes, Slurm, or 20+ listed clouds without code changes
- Auto-failover on capacity errors and cheapest-available placement across infra
- Autostop of idle clusters and binpacking on shared clusters reduce idle GPU spend
- Runs inside your own cloud accounts, VPCs and clusters (BYOC)
Weaknesses
- Infra orchestrator, not a model library or trainer; you bring your own training code
- Needs cloud credentials and GPU quota set up per provider before use
- Sandboxes feature is early access via a signup form
- Team deployment requires running a separate API server
What it needs
- GPU optional
- Docker
- Needs Kubernetes, Slurm or a cloud account (AWS, GCP, Azure, etc.)
- Models: vLLM, SGLang, Ollama
Quick answers
Is skypilot open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Can you run skypilot with Docker?
Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.
Does skypilot need a GPU?
No. It runs on a CPU and can use a GPU when you have one.
Also in ML frameworks and libraries
See all 168| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd | 93 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and training scripts | 93 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 92 out of 100 |