I like how Modal Labs completely eliminates the infrastructure tax since we don't have to manage Kubernetes, configure CUDA drivers, or manually build Docker containers just to run a Python script on a GPU. Their custom container runtime boots up in seconds, allowing us to run local code on cloud GPUs almost instantly. I appreciate defining the GPU environment directly in Python with simple decorators, avoiding the headache of Docker, Kubernetes, and CUDA. I also enjoy Modal's shared storage volumes and network file systems for machine learning, which let us persist heavy model weights and training datasets across separate runs. The parallel mapping feature is particularly beneficial, enabling us to effortlessly scale a single Python function across hundreds of parallel CPU or GPU containers to process large datasets in minutes rather than hours. Additionally, the initial setup was straightforward with just pip commands.

Solid offering, a lot of unexpected payments that lead us to switch to another provider

I really appreciate the main dashboard's layout in Modal Labs because it has the very bare minimums needed to get the job done without distractions from unnecessary decorative elements. I also like the eco-friendly dark theme. The initial setup of Modal Labs was very, very easy for me, which was a big plus.