Model flexibility

DataQI is designed to operate securely regardless of your infrastructure’s network constraints. It supports a flexible model architecture, allowing you to choose between remote, cloud-hosted Large Language Models (LLMs) and locally hosted models running on your own hardware.

Remote and SaaS models

For environments with standard network egress capabilities, DataQI can connect securely to enterprise-grade, cloud-hosted AI providers (such as Azure OpenAI, Google Gemini, or Anthropic). This approach typically offers the highest reasoning capabilities and requires no local GPU infrastructure.

Local and air-gapped models

If your organisation operates in a strictly air-gapped environment or requires absolute data sovereignty, DataQI supports locally hosted models. By connecting to a local model gateway (such as Ollama or vLLM), DataQI can run entirely within your secure network perimeter using local GPU hardware. No data ever leaves your internal infrastructure.