Paperspace Reviews

I placed an order for some clothing items on January 2, 2023, but I still haven't gotten any of my stuff. I've also asked for a refund, but there's been no reply. Please, I really need my refund ASAP. Thanks!

As a student in computer science focusing on machine learning, I thought paperspace gradient would be the perfect platform for training and developing ML models in the cloud. Unfortunately, this service seems to miss the mark on almost every level.

To start with, the user account management is a hassle. You can’t just delete your credit card info; you can only swap it out. Plus, there’s no way to remove your account through the interface, which is pretty annoying. So, is my card info just stuck there forever? Other reviews have already covered the confusing hidden fees (like paying monthly just to pay more for GPU access), so I won’t dwell on that. They’ve also introduced force-stop limits for notebooks, which is something, I guess.

The product has mostly regressed. They’ve added features like automatic model deployment and dataset versioning, but they don’t function properly. I’ve never been able to upload a dataset using their system—not in 2021, not now. It fails through the CLI and the web UI. The cross-notebook persistent storage feature, which is actually useful, is buried deep in the documentation and lacks both CLI and web UI upload options. Thanks for that.

And what’s the deal with the CLI? You don’t need to try to compete with AWS. Just make a decent web interface. I get that CLIs are trendy, but the web UI barely works as it is.

The main gradient product still doesn’t function well, with notebooks crashing randomly or browser tabs freezing. Plus, unlike in 2020/2021, most of the free GPUs are usually taken, which kinda defeats the whole point of having them.

I also found it amusing that back in 2020, they bombarded me with around 30 emails asking about my experience and if I wanted to chat with the dev team. Those emails stopped right after I agreed to a free month of gradient pro. I guess $8 is where they draw the line for customer engagement. Fair enough, lol.

In short, I really wish this product worked better. If I could just upload my dataset to a reliable persistent storage (without having to fork out more cash for an S3 container) and get a Jupyter notebook with a GPU up and running in a few minutes, I’d give this thing 7 stars. Instead, it’s just a frustrating experience every time I try to use it, and I’m left hoping Google Colab eventually gets better than this GPU roulette.

While I was trying to get their pro plan, my card got declined. What’s up with that? They didn’t bother to hide the error logs, and it looks like they attempted to charge me a grand directly to Strapi instead of just 8 bucks. I'm seriously ticked off, feels like a total scam.

Paperspace has truly been a lifesaver for my job. I frequently need to train several machine learning models, and if the supercomputer at my workplace fails, I can always rely on Paperspace to offer a user-friendly and accessible service to keep my projects on track.

As a research assistant, I don’t use tools too heavily. I recently signed up for the 'Pro' plan, which is a mid-tier option. It provides a GPU that's an upgrade from the Free plan. Up to now, I’m quite pleased with this plan and the service overall. However, I do have a concern regarding the utility; it’s about how many notebooks I can operate at once. Paperspace’s Pro Plan allows for 3 notebooks to run at the same time, but I can only manage to run 1 notebook for some reason.

I gotta say, the service is top-notch. They offer Jupyter notebooks on some seriously powerful machines, and the prices are pretty reasonable. Plus, there are even free options if you’re willing to pay a small monthly fee.

Paperspace is truly the top cloud GPU service for machine learning students such as me. It handled three of my personal projects effortlessly, offering unlimited GPU time for only $8 each month. I plan to stick with Paperspace unless they make some outrageous changes to their pricing.

As a data science student and a newbie researcher, I stumbled upon Paperspace while searching for solid computing power for accurate pricing. I explored various cloud computing options, but they were just too pricey for small to medium-sized projects. The free or cheaper alternatives had so many restrictions that I couldn't run any advanced deep learning models due to running out of RAM in no time. With Paperspace, that's not a concern. The pricing is quite reasonable for the value you receive, plus you can deploy models without needing to set up a formal API on another hosting platform. It's a one-stop shop for both researchers and professionals. The gradient UI is super user-friendly and provides useful metrics on energy and memory usage, helping you keep tabs on your model's performance or any application you're working on. Their pro version has a fantastic selection of GPUs, and you can access some powerful options for just $8 a month.
I wholeheartedly endorse Paperspace as the go-to machine learning platform for anyone eager to dive deep into deep learning or use machine learning in a professional capacity.

I ain't no machine learning whiz or data guru - I’m in the media biz - but Paperspace has been a game changer for me. I've had zero problems with uptime or performance, and the desktop streaming is super smooth with hardly any lag. There are plenty of ways to create the perfect setup, featuring all the usual tools I expect, plus some cool extras I didn’t even know I needed, and their data centers are solid on both U.S. coasts and in Europe.

The Customer Support and Customer Success teams are really quick to respond and super helpful. I've had nothing but positive experiences with them, and I even get to tap into their experts to fine-tune my setup to fit my needs.

I’ve seen a bunch of reviews talking about billing and overages; since I’m a CORE user, I can’t speak to how it works on the Gradient side (both accounts have access), but you can set a time limit for the machines so they automatically shut down after a certain number of hours. So far, it hasn’t let me down.

There’s also a global compute limit feature, allowing you to set a max spend and an alert for when you’re nearing that limit. I’m not sure if these are new, but you can definitely set up some boundaries to avoid any surprises on your bill.

I use my machines a ton, and I’ve never felt the need to set a compute limit. Your experience might differ, and sure, you could probably find something cheaper, but for the mix of quality and ease of use, Paperspace is a stellar option.

One last thing: Paperspace seems to be improving all the time, so many of the issues mentioned in reviews have been sorted out. From what I’ve seen, they really prioritize usability and are quick to enhance their product, faster than I’ve noticed with other solutions.

Pros: Quick and user-friendly.

Cons:

1. It’d be great if Paperspace included a tech solution to avoid the hassle of reinstalling the same package repeatedly. Otherwise, it’s just a waste of my time and the electricity from Paperspace...

2. For machines that charge by the hour, it would be helpful to get reminders for active users so they don’t accidentally leave them running. I ended up with a $50 bill from Paperspace just because I forgot to switch one off...

I've been utilizing Paperspace to dive into the fastai textbook, and I run the fastai codes through Jupyter Notebook. Honestly, it's been a fantastic experience so far. No problems at all, and everything runs like a charm.

The design is simple and user-friendly. There are plenty of useful videos to help kick off a project. They manage the fast.ai book way better than Google colab. However, they lack the ability to start a project without first choosing a project style (like fast.ai, R, visual analytics, etc.) that automatically fills in some code in the project container.

I've been using Paperspace for over six months now, and I have some mixed feelings about their offerings. At first, I was on the free plan, which was decent; you get a K80 GPU and 5GB of storage, and I had no complaints there. Eventually, I decided to upgrade to the Pro plan, which costs around $8 monthly. This plan gives you access to bigger GPUs like the RTX4000, RTX5000, P5000, and offers 15GB of storage. Everything was going well until I checked my bill a month later and saw it had jumped to about $35, and it just kept climbing. I hadn’t really paid attention to the bills until the fourth month, and by then, I had already lost over $130, which feels way too steep for what they offer. The extra charges came from exceeding the 15GB storage limit; I ended up using over 50GB without realizing it. What really bugs me is that I wish I had been notified about the extra storage usage via email or some kind of alert instead of just getting charged directly (I’m a student, after all). All I want is better communication before they start charging my card without a heads-up. On the bright side, their customer service is top-notch; they respond quickly, which is a plus.

Awesome support and service! I ran into a few issues with my machine, but they sorted it out fast and even credited me for the downtime. Really impressed with the service, can't wait to host more VMs with them.

The service is alright, but there are some sneaky charges, and their billing system has some glitches... If I had known better, I would’ve picked a different option from the get-go, so I ended up wasting both my cash and my time.

Update: customer service finally got back to me regarding the unexpected fees, but they couldn’t provide a clear breakdown of what I was charged. It looks like even they don’t really understand how the costs are figured out. It’s nice to have someone to chat with, even if it didn’t help much.