Runpod Inc Recenzje 214

They estimated I have 50 minutes left in my pod, I needed 10 minutes to finish a Lora Training. After 2 minutes my pod was terminated. I wasted 10 hours of my life and $30. Pure scammers. Also, they never gave the first deposit bonus they promise if you come from affiliate links!!! STAY AWAY!

Runpod surfe sur la vague de l'IA et a flairé le bon filon du "GPU as a service". Malheureusement, le service n'est pas à la hauteur et les pratiques de facturation assez douteuses :
- on vous avertit qu'un pod en cours doit être "arrêté pour maintenance", vous devez tout recommencer - à vos frais bien sûr
- vous êtes facturé pendant que le pod se met en place, même si ça dure une heure (et que ça n'aboutit jamais, ce qui arrive de temps en temps)
- un pod peut continuer à être facturé après sa suppression et grignotter tout votre solde en une journée. Bon courage pour leur faire admettre. C'est d'ailleurs la raison de ma résiliation.
- service client incompétent et obtus, incapable de produire des preuves de ce qu'il avance, et qui modifie ses réponses à mesure que vous les mettez devant leurs contradictions.
Une seule chose intéresse Runpod : votre argent.
Je ne comprends pas qu'ils soient autorisés dans l'Union Européenne.
Fuyez !

Really good

Absolute 10 out 10 experience the set up is easy they have so many gpu options and they are super fast and work amazing and its such a goood deal for what im getting absolutely amazing

Overall the experience is great but the API is not that great to get costs from projects

I had a very positive experience with Runpod. What really stood out to me was their responsiveness and professionalism it’s clear they genuinely care about their users and take feedback seriously.

A special thanks to Luke for his kind and professional approach. His response and support reflected a true customer-first mindset, which is something not easy to find these days.

In addition, the platform itself is very well designed easy to use, flexible, and suitable for different workloads. The overall experience feels smooth, and you can see the effort put into making things efficient for users.

The combination of a strong product and a professional support team shows great potential for the platform’s future growth.

I would definitely recommend Runpod to others, and I’ll be sharing my experience in both English and Arabic with my network.

No disk (pod) could be installed. No template works. The AI ​​assistant helps you by actually doing everything yourself. Everything is wrong. You need to be a programmer to work here. I think this is a big scam. I think the authorities should deal with this platform, because this is a classic internet scam that wastes your time and money. I have been trying to configure the disk for 30 hours and have not succeeded with the help of all resources. Skip this if you don't want to be in hell.

0 customer support = 1 star, spent around 600k USD and the support still completely ignoring my messages, making millions but not able to help customers is a shame.

Brand new to the platform and I had a wonderful experience!

Sorry, but it was hit and miss. We'll likely reconsider our plans and possibly switch to a different cloud provider for our project as the experience was terrible.
It is quite a common problem for pods to have very slow network speeds, causing long pod start ups (>20 minutes) despite small Docker images (<2 GB).
Sometimes, the spawned GPU pods didn't have OPTIX available, sometimes they didn't even have CUDA! So, basically broken pods.
And these were "Secure cloud" pods.
I hardly imagine how this quality can be considered acceptable for a cloud provider.

Terrible, terrible service. Servers are now hardly ever available, even less so if you want to attach a permanent space disk to one. Can spend hours trying to find a server that works -then it might take hours to initiate (after uploading loads of data to it), and then crash again, meaning you have to restart. They create pods and storage areas on servers they know are broken. Customer service is basically non-existent, only a useless chatbot. Even if your server is down and you're promised not to be charged, they keep on charging you money. Truly a scam, should be shut down.

Really great experience. I do NOT know coding at all. I'm using Claude Code to guide me through everything on Runpod. Set up is quick/painless. Price is very reasonable.

Happy customer, great, constant innovation. Excellent customer support.

Extremely confusing. Difficult to setup. Very poor experience.

I’ve been trying for several days to use RTX 6000 Pro for WAN Animate, and I’m already down about $20 with almost nothing to show for it. A week ago, when I first started using the service, everything worked fast and smoothly. Now pods take up to 30 minutes just to load . if they load at all. In many cases, they fail to initialize, yet I still get charged for the time.

It honestly feels like the service has been deliberately slowed down while billing continues as usual. Support is completely silent.

I’ll use my remaining $7 balance and that’s it. Not recommended.

I'm not sure if this isn't a scam. I tried a qwen 2.5 model for an hour asking a couple of questions. Mainly trying to determine if I could get an answer from a prompt. Then I tried qwen 3 80b. In that period I had the remainder of my $10 taken and I had 0 questions answered. The machines all failed to initialise but I got charged for it. Using chat gpt/claude you pay $15 and get a whole months use. Not 2 hours of use. Don't bother.

world best cloud GPU platform
I love Runpod

I would only use runpod, if using the community cloud platform. Their secure cloud platform uses archaic CPUs with very low clock speed that practically makes most training much slower for mixed workflows. Additionally their CPU only pods are ridiculously outdated. I tried using their "CPU5" level pod, which is supposed to be 5GHZ of high performance usage. I received a heavily outdated AMD EPYC 4564P, which performed terribly.

Overall, RunPod delivers a strong service with dependable infrastructure, and their support is consistently fast and effective. The main drawback is limited availability of certain community GPU instances (e.g., A6000) as well as occasional issues with network performance.