I'm a data science student and first-steps researcher. I ran into paperspace because I needed decent computing power for proper pricing. I tried all cloud computing solutions, but they were expensive for small to mid-size solutions. Some of the cheapest/free ones have very restricted limitations, and I couldn't implement any DL advanced model or architecture because I ran out of RAM very quickly. With paperspace you don't have to worry about that. It offers very fair pricing for what you get in exchange, you can also deploy models, and you don't depend on building any formal API on any other hosting service. It's an all-at-once for researchers or professionals. Its gradient UI is very intuitive and informative, for it displays metrics about energy and memory consumption of your hardware resources to efficiently track your model's performance or any application you are running in this environment. Its pro version offers a great variety of GPUs, and you can have access to powerful ones for only $8 a month. I highly recommend paperspace as a default machine learning platform for anyone wanting to deep dive into deep learning or use machine learning professionally.

I like their GPU-based severs, making it very easy to deploy a GPU-based server in minutes.

What I like about Paperspace is it provides Virtual Machines fast and cheap. Its UI is simple and beautiful. It offers a lot of machines from Standard GPU, CPU only, and with CUDA.