RunpodReviews of runpod.io

I feel like I've been scammed... I had a small, multi-day PyTorch job I was hoping to run, so I paid $20 in credit. I selected an RTX 3090 at $0.50/hour, enabled permanent storage (of which I used 9 GB at $0.09/GB), This should therefore have lasted around 2 days. And left it running for 16 hours; I came back, and it said I was in debt -$4.90?! I checked my billing stats; I've spent $0.07 on storage, $8.11 on pods - So RunPod, where is the rest of my money?! I feel I have been stolen from. I then had to pay another $10 just to get my data back, since I was $4.90 in debt. Terrible experience. (Not to mention, it took ages to upload and download files, wasting paid-for time, nor can I seem to access my data without another pod - so if you run out of credit you'll at minimum have to pay $10 to get your data)

such a complicated and not even worth to buy

A bit pricey. GPU availibility close to none. Internet that is often slowed down to the minimum to have you pay almost an hour just to launch a machine. Extremely slow docker pull, no cache ofc. You can't lock GPUs anymore (but you can still keep paying for it to be "locked" !) (makes sense as they have no GPU available). Hazardous hardware. In the "security" cloud part, where it is sold as "everything should be stable and same everywhere", well some GPU have update firmware, some don't. Had better experience to deploy their official docker images on "community" cloud. I mean, if your official docker images don't run on your own hardware. You are doing it wrong. This is beyond stupid.

Engineers using RunPod: check your GPU clocks and training throughput before leaving a job running overnight. My experience shows how expensive severely underclocked compute can become. I originally liked RunPod. It felt accessible and friendly for personal AI projects. The handling of this incident has changed that. From September 23–26, my RTX PRO 6000 Blackwell Workstation GPU delivered about 21% of my reference run’s training throughput, while billing continued at the full hourly rate. Same GPU model, matching core training code and model settings: - Median GPU clock: 2,662 MHz → 585 MHz - Training throughput: 645 → 137 samples/second - Both reported roughly 600 W under load The second dataset was larger; the comparison above is per sample. The slowdown persisted across thousands of telemetry readings. On October 2, RunPod support confirmed low clocks on the host under a full 600 W load and repeated NVIDIA driver errors. The precise cause remains unresolved. A moderator in RunPod’s Discord also acknowledged that more than one user had raised concerns about this issue. As of October 9, I have not received a clear root-cause explanation or recurrence-prevention plan. The dashboard I checked did not display GPU clock frequency. I also have no commitment to a clock-frequency dashboard or automatic underclocking alerts. Basic clock visibility should be the minimum. The advice was to spot the problem myself, stop the Pod and deploy elsewhere. Customers are responsible for their code. Monitoring host health and addressing a documented host-side performance problem should be the provider’s responsibility. Engineers should be able to leave training running overnight without babysitting rented hardware to protect themselves from wasted charges. I asked for a proportional refund to my original payment method. On October 8, RunPod refused, citing its policy on consumed services. Its email described the credit adjustment as a one-time exception, with no guarantee of similar adjustments in future incidents. Credits requiring me to keep using the platform are not an adequate remedy for me. Framing this as an exceptional courtesy, while leaving the same risk with the customer, is unacceptable. Imagine renting a supercar for the Nürburgring and discovering it delivers only a quarter of its normal power. The rental company refuses a refund because you already drove it, then offers credit toward another rental. That is how this feels. In my view, a “consumed usage” policy should not shield a provider from responsibility for defective service. I want a clear technical explanation, safeguards and a proportionate cash remedy. Until then, do not assume high GPU utilization means you are receiving the performance you paid for.

- Slowest support possible (36-72 hours to respond to ticket while pod is being billed by millisecond? Yeah, sure guys, i'll wait. Response was useless btw.) - Near ZERO GPU supply - Broken billing & server stats (you can pay for 1TB of ram but receive only 300) - Can get totaly bricked server. (Just a roulette.. you can get server with broken nvswitch, locked gpus, less ram or zero mounted disk space)

I had never heard of RunPod and was never a customer of theirs. My debit card was used by someone else to open a fraudulent RunPod account, resulting in $810 in unauthorized charges over two weeks. I reported this to RunPod immediately with clear evidence: the account was created using an email address I have never used or owned, one that has never been associated with any of my real accounts, and the account creation IP address traced to a Florida city I have no connection to. RunPod confirmed the account was fraudulent and banned it. Despite this, RunPod refused to refund the charges, stating they found no evidence of third-party access – even after banning the account for exactly that reason. Their response also suggested I should have caught the fraud sooner by checking my bank statement daily, rather than acknowledging their own system’s failure: the account was opened with an entirely different email than the one tied to my payment method, and no purchase or account notifications were ever sent to my real email. That mismatch should have been flagged on their end, not left for someone who was never even a RunPod customer to discover after $810 in charges. I want to warn other potential customers: RunPod’s stated fraud protections did not function as expected, and when presented with concrete evidence (mismatched account email, verified IP location discrepancy), their team dismissed it and shifted responsibility onto me rather than acknowledging their own account verification process failed. I’ve had to pursue this through my bank’s fraud dispute process instead, since RunPod would not resolve it directly despite acknowledging the account was fraudulent. If you’re considering RunPod, be aware of how they handle confirmed fraud on their platform – even when the victim was never their customer to begin with.

It is becoming unusable to the unability to rent 5090 that all have broken templates and cuda libraries missing. The support is very bad not understanding your problems and not refunding money for pods that where not working after you showed them problems with bandwith or performarce. Better find other services

You took my money and said GPU's are not available bro seriously? I am frustrated from 3 days from this problem whenever I use serverless after one request on avg have to wait atleast 30 mintutes to get another gpu. But Even after gettting hardware until my model loads into it you took it away and call it throatling while draining my balance... Why I paid you. If you don't have enough resources stop promoting your shit... and stop taking money when their is no availability.. Runpod is not worth to build anything production ready... I wish I could gave you less than 1 Star