Thank you for your explanation. I really appreciate your honesty.
The following lines are based on my personal experience regarding Floydhub services. I hope my comment could help you to improve the future user's experience.
I introduce briefly myself. I am a student passionated by the field of deep learning and I really enjoy making A.I. application/ tutorial which I make available later on my github. By the way, most of my project are made using floydhub and I advise in my tutorial to use Floydhub for training (for example https://github.com/OValery16/swap-face).
I think you did a wonderful effort in term of design, userfriendlyness and flexibily. Few month ago, I got a problem with the training dataset offloading speed, and you solved this problem in almost no time. I really appreciate the reactivity of your team.
However, as you can imagine, I am a study and the financial aspect is extremly important for me. I understand your point of view about the SLA problem. But I have to tell you that many users (especially students) won't be able to pay $1.20 per hour for using you GPU, and will endup leaving floydhub for other cloud platforms such as Google Cloud Service, crestle ...
I don't critize your service. I just enjoy doing deep learning applications, because I have a stong passion for DL. I would like to be able to continue what I like using your service.
Can think about making some offerts specific for student ? (for student the main point is not the SLA, but the price of the cloud service)