HTTPS/TLS encryption between your application, QDivZero and GPU infrastructure, with zero logs of inference content and no data sent to model creators.
Infrastructure and public inference providers
We are part of the NVIDIA Inception program.
Private deployments rent GPUs on demand from Runpod and Vast.ai. Public models are served through Azure, GreenPT, OVH and Cerebras. Those providers process requests to run the models. QDivZero manages access and applies its privacy policy. Unlimited runs in Europe. Other access options depend on their infrastructure or selected region.
Prompts, files and responses travel over HTTPS/TLS between your application and QDivZero, and between QDivZero and the GPU or public inference provider. Encryption protects traffic on both the request and response paths.
We apply zero logs to inference content, except for context retained by the Responses API to continue conversations. Under its privacy policy, QDivZero does not access that context or use it for other purposes. Your content is not used to train models or sent to their creators. You can delete logs at any time. If you disable them globally, use Chat Completions.
The selected region determines where your model runs. Our policy of zero logs and no data sent to model creators applies to every deployment, regardless of the GPU provider or region.
The Compute provider network brings together established cloud operators, GPU-native specialists, and regional infrastructure teams. Each contributes a different mix of hardware, locations, and available capacity.
Providers with mature regional footprints, networking, and operational processes for broad infrastructure coverage.
Operators focused on accelerator infrastructure, with deep hardware catalogs and high-density GPU capacity.
Infrastructure teams with local availability and clear data-center locations for geographic requirements.
Providers able to maintain availability, incident response, and dependable provisioning as capacity changes.
Joining the network is more than listing GPUs. We review the infrastructure and operating model so the capacity presented in Compute is clear, supportable, and ready for ongoing operation.
GPU models, VRAM, topology, and available capacity.
Data-center locations and regional operating coverage.
Networking, launch processes, monitoring, and support.
Availability updates, incident handling, and operational follow-through.
Operate cloud, GPU, or data-center capacity? Tell us about your inventory, regions, and operating model.
