GPUHub is still beta on nodepedia. The tagline that fits the scrape is consumer and prosumer NVIDIA rentals: the published rows include RTX 4080, RTX 4090, RTX 5090, and an A800. That is a real list, not an empty profile. What is still missing is everything around the cards.
Why GPUHub stands out
There is no confirmed platform story yet. API, Docker, Jupyter, Kubernetes, persistent storage, multi-GPU, spot, and reserved are all unset. Billing unit is unknown. No payment method is recorded. No founding year and no headquarters. Ease of use and enterprise readiness in our scores are at the floor, which matches “we could scrape a price table and almost nothing else.”
Read the GPU table. Do not read this page as a recommendation to move a workload here.
Pros
- A scraped consumer and prosumer list, including RTX 4090 and RTX 5090
- Still being checked, so a later audit can promote it if the missing fields get filled in
Cons
- Beta, and this page stays a draft until an audit says otherwise
- Billing unit and payment methods are unknown
- No recorded API, Docker, Jupyter, storage, or multi-GPU support. Absence here means unconfirmed, not a lab test that proved the feature missing
- No founding details on file
Getting started
- Open GPUHub and check the live card list against the rows on this page.
- Confirm billing unit and how you pay before starting anything that runs long.
- Confirm the access method yourself. We have not recorded Docker, Jupyter, or an API.
- Keep the job small until the provider leaves beta here.
Best for: A short test of a consumer NVIDIA card, after you confirm billing on their site. Not a place to park a production job on the strength of this profile.