You are billed per millisecond of compute. We charge $0.00055 per second of compute regardless if its a CPU or GPU based model.
How do we handle cold starts?
We have setup our infrastructure in such a way that our cold starts are less than 1 second
What GPU's do we use?
Currently we only support Nvidia A16's or V100S SXM2. We are looking to add A100 in the coming months
Can I run my models on my own infrastructure?
Unfortunately we do not currently support running our platform on your own infrastructure. If this is a really important requirement for you, please reach out to email@example.com and we can see if we can help you.
Is Cerebrium secure? Where are my models hosted?
Yes. Cerebrium is GDPR and SOC 2 Type I compliant. All models are hosted within our infrastructure and all API calls made to Cerebrium are SSL encrypted.
Ready to deploy your models in less than 5 minutes?