Scaling the data platform powering ChatGPT

How do Databricks and OpenAI bring governed frontier models to the enterprise?

The operational trust between the two companies lays the foundation for a larger commercial partnership. Through a landmark multi-year agreement, Databricks and OpenAI have partnered to make OpenAI's frontier models, including the latest GPT models and Codex, available natively inside the Databricks Platform. Enterprises can now use those models without moving sensitive data, invoking them via SQL or API calls against data already under Unity Catalog governance.

The native availability pairs with Agent Bricks, Databricks' framework for building, evaluating, and scaling production-ready AI agents. Agent Bricks measures accuracy and helps optimize models like GPT-5.6 for domain-specific tasks, while Unity Catalog provides the access controls, lineage, and audit trail needed for enterprise deployment. Together, the integrations mean customers no longer have to choose between frontier model quality and enterprise-grade governance.

Greg Ulrich, Chief AI and Data Officer at Mastercard, said the combination unlocked something specific for his team: "For any use case, AI agents come down to three things: quality, scale, and trust. This partnership between Databricks and OpenAI provides us the opportunity to build trusted AI agents." The partnership extends what OpenAI is already proving at its own scale: that frontier-AI workloads can run on a governed, lakehouse-based foundation, available to every Databricks customer.