Runpod Reviews 

RunPod works perfectly. Pay-per-second, great GPU selection, no hidden fees.

The negative reviews? Users who didn't read the docs or left their pods running for days wondering why they got charged.

Learn how ephemeral storage works, set spending limits, and you'll be fine. Not a scam – just user error.

Prices are increasing while availability is decreasing. And the site still needs work. It shows you available GPU that aren't.

Overall good experience. User-friendly interface. Clear posting of prices. Positive interaction with customer service. Responsive, courteous, and action-oriented to assist with the actual inquiries and concerns.

The existing pod couldn’t even start because there’s no GPU. You have to delete it and rebuild with a different setup.
Download speeds are deliberately throttled — we’re talking just a few MB/s for a model.

Classic business tactic.
From my experience, teams like this tend to be short‑sighted, and I seriously question whether data stored on their platform is safe.

Overall, the experience is poor. Definitely not recommended.

It was really good initially, but lately none of the cheap community or security pods available, I subscribed to runpods because they have plenty of good pods available most of the times like less that $0.19 per hr. But looks like thats not the case anymore.

Pods always getting killed. Unreliable

I've tried a few of these pod providers, I shan't name them because I don't think that's fair but this is by far the best so far. It's reliable, and when a gpu falls from availabilty it rebuilds it somewhere else easily, which is more than i can say for some others !
It's a little pricey for people like me who are just dabbling and trying to build a small second income, but I guess in the end you get what you pay for !

Assistenza veloce e disponibile

The service is not super consistent, prices are creeping up, and the quality is going down. They have been talking about network drives that could deploy on any pod in any region since I joined but its still not a reality. The pods get bogged down when there is heavy traffic in a region. Idk. Some days there is no gpu availability for hours at a time. Known templates fail to deploy, even though the templates had not been changed since the last successful deployment. Uploads and downloads can be as low as 30kbps... making a seevice useless where loading a 30 gig model is common. They are making changes to deployment pages instead of addressing these issues.

I would give zero stars if I could. I just spent over 50 hours and a pile of money trying to get a basic AI agent setup (OpenClaw) working with "persistent storage" on RunPod.

The marketing says Network Volumes give you reliable persistence.

The reality? A pathetic joke. Files kept installing to ephemeral locations that get wiped every single restart, no matter how many symlinks, PATH hacks, or support tickets I tried. I lost all progress multiple times, had to reinstall everything from scratch repeatedly, and spent hours manually typing long API keys because their web console is garbage.

Their own support team eventually just shrugged and said "try creating a new pod lmfao."

Meanwhile, I moved to Paperspace and had a fully stable, persistent setup in under 1 hour.

If you enjoy burning massive amounts of time and money on a platform that can't even deliver basic persistence correctly, RunPod is perfect for you. If you actually want your data and configuration to survive a simple restart like a normal product should, run away as fast as you can.

Save yourself the headache. RunPod is great for disposable GPU experiments. For anything serious that needs to actually persist? Absolute trash.

Very convenient, simple, rich.
Only issue is gpu availability when there is huge demand.
Please expand in north Africa :)

Swift, collaborative, precise. Thank you for continuously providing a State of the Art product and supporting your customers.

Terrible, serverless workers are constantly throttled

Very bad experience, my server got reseted twice when training a model because of a setting change in the run pod panel that made it crash. Do not recommend, I thought paying more for a more professional service wasn't worth but it actually is.

We’ve been running a few serverless endpoints that were deployed about a month ago. Everything was stable and working as expected — no changes, no redeployments, no config updates.

Suddenly, we saw a sharp spike in billing over a short period. After reporting the issue, things returned to normal without us doing anything.

Instead of addressing the issue, they’re offering small credits, which don’t address the core issue.

I've tried several GPU rental providers but Runpod stands out - for the UX of their console and their great community. They also offer loads of useful tools and have almost perfect documentation if you are the kind to prefer self-help like me.

Honestly, it seems like the era of runpod being a good product is over. In the matter of a few tens of hours, they raised spot prices by 25%, then removed them entirely, then brought them back gated behind API. The spot functionality wasn't technical debt nor troublesome for the web UI; it was a simple plug and play with other HTML elements already on dashboards. This move then can be seen as nothing more than an choice to intentionally add layers of friction and abstraction so as to reduce the value of their product to their consumer base. It comes with a rotten smell for company incentives.