If you need a cloud computing service that offers one-click access to Jupyter notebooks and supports multi-GPU instances for AI work, Lambda is a strong contender. Lambda offers on-demand GPU clusters with multi-GPU instances, including 1x, 2x, 4x and 8x NVIDIA GPU configurations. It also offers one-click access to Jupyter notebooks and preconfigured ML environments with popular frameworks like TensorFlow and PyTorch. The service supports a range of GPUs, including NVIDIA H100, H200 and GH200 Tensor Core GPUs, and charges by the second.
Another option is RunPod, a globally distributed GPU cloud service that lets you spin up GPU pods instantly. It supports a range of GPU workloads and offers serverless ML inference with autoscaling and job queuing. RunPod also offers instant hot-reloading for local changes and more than 50 preconfigured templates for frameworks like PyTorch and Tensorflow. The service offers a CLI tool for easy provisioning and deployment, and it charges by the type of GPU instance and usage.
For a more open-source approach, dstack is worth a look. dstack automates infrastructure provisioning for the development, training and deployment of AI models on a range of cloud services, including AWS, GCP, Azure and others. It makes it easier to set up and run AI workloads, so you can concentrate on your data and research while taking advantage of cheap cloud GPUs. dstack supports a range of deployment options and offers extensive documentation and community-driven support.