Native OpenAI Models Now Generally Available on Databricks

OpenAI models, including GPT-5, are now generally available on the Databricks Data Intelligence Platform. This launch is a key part of our Week of Agents, showcasing new releases to help you build, govern, and deploy powerful AI agents securely at scale.

With GA, OpenAI models on Databricks reach full production readiness. This marks another step toward Databricks’ mission to make leading foundation models available natively within a single, governed platform so enterprises can build high-quality agents that reason over their data securely and at scale.

Build with OpenAI, Where Your Data Lives

Now, enterprises can bring OpenAI’s frontier intelligence directly to their governed data, without moving data around. This integration enables systems that not only generate language but also understand, reason, and act on enterprise data in real time.

Teams are already using this combination to:

• Enrich millions of support tickets or product reviews with automated insights
• Generate structured summaries and action items from meeting transcripts or contracts
• Build conversational agents that answer questions grounded in live Delta data

What You Can Do with OpenAI Models on Databricks?

Explore Models in the AI Playground

The AI Playground offers the fastest path to start experimenting. Open the Playground tab, choose an OpenAI model, and compare GPT-5, GPT-mini, and GPT-nano side by side. Adjust temperature, max tokens, and other parameters interactively, then export the best prompt directly into a notebook or SQL cell.

Apply OpenAI Models to Your Data

OpenAI models can be used like a built-in operator in SQL or Python, making it simple to analyze contracts, PDFs, transcripts, or images directly where your data lives. When you run these queries, Databricks automatically scales OpenAI model capacity in the backend to handle everything from a handful of rows to millions, ensuring fast, reliable results without extra setup.

Automate Document Intelligence

OpenAI models can be chained into Lakeflow Declarative Pipelines to build GenAI ETL, ingesting data from systems like Salesforce or Dynamics 365, applying AI transformations such as summarization or classification, and orchestrating the flow with built-in governance and observability. This pattern scales to millions of documents, creating end-to-end document-intelligence pipelines with lineage, auditability, and quality monitoring built in.

Build Intelligent Agents with OpenAI and Agent Bricks

Agent Bricks turns OpenAI models into intelligent agents by connecting them to enterprise data in the Lakehouse, vectorized knowledge through Vector Search, and external systems via MCP. 

Once connected, Agent Bricks automates prompt and model optimization, selecting the best combination for each use case and continuously improving performance through built-in evaluation, observability, and governance. 

When ready, agents can be securely deployed at scale, fully governed by Unity Catalog and monitored through AI Gateway.

Govern and Monitor with AI Gateway

All OpenAI traffic on Databricks is managed through AI Gateway, the unified control plane for enterprise AI workloads. Every GPT-5, GPT-mini, and GPT-nano request is governed, logged, and auditable — fully integrated with Unity Catalog. Teams gain visibility into usage and token spend, can enforce rate limits, and enable fallbacks to make API calls more reliable.

Start Building Today

• Join us on 11/11/25 for “The Future of AI: Build Agents that Work” with Sam Altman and Ali Ghodsi.
• Try OpenAI GPT-5 in the AI playground.
• To go deeper, explore the Foundation Model API documentation to learn how to integrate OpenAI into your data and AI workflows.