Pay for what you run.
Public models with published inference rates
Use pre-deployed models without selecting or running a GPU. Rates are shown in EUR per 1 million input and output tokens as a standard usage reference, and charges are calculated from recorded inference usage.
Read the Public Models guidePublic Models use metered inference pricing at the published input and output rates. Compute below is separate and bills active GPU time.
Estimate GPU cost for a model
Search for a Hugging Face model to estimate the VRAM it needs and see compatible infrastructure at current GPU-hour prices. This is a GPU compute estimate, not token or API pricing for the model.
Model-based GPU estimate
Choose a model or quantized GGUF variant to estimate VRAM and GPU cost
Browse GPU prices directly
Already know the GPU or VRAM you need? Browse this inventory independently of the model estimate above. Compare live capacity, location, market type, and the infrastructure price per GPU hour.
All GPU configurations
Live infrastructure price per GPU hour
1796 configurations
All GPU configurations
Live infrastructure price per GPU hour
€0.02/h
Limited capacity
€0.03/h
Limited capacity
€0.04/h
Limited capacity
€0.04/h
Limited capacity
€0.04/h
Limited capacity
€0.04/h
Limited capacity
€0.04/h
Limited capacity
€0.04/h
Limited capacity
€0.05/h
Limited capacity
€0.05/h
Limited capacity
Showing 1-10 of 1690
What your deployment can include
Firewall and Smart Balancers are included with every deployment, so protection and resilient routing do not add a separate platform fee. Add Flexible Vector Database only when your application needs managed semantic retrieval.
Flexible Vector Database
Vector storage billed by GB-hour
€0.0010/GB/h
Firewall
Included
Smart Balancers
Included
Prices updated 9/7/2026, 6:03:05 AM
How billing works
Review how QDivZero records debits, balances, top-ups, and invoices across your account.
Read the billing guideReady to run a model?
Compare GPU capacity, choose the configuration your workload needs, and deploy on QDivZero.