OpenAI GPT-5.5 Hits Databricks

OpenAI's latest frontier model, GPT-5.5, along with its coding assistant Codex, is now natively available on the Databricks Lakehouse Platform. This move integrates advanced AI capabilities directly into enterprise data workflows.

The partnership aims to bring powerful AI tools to businesses with a focus on governance and security. All usage of GPT-5.5 and Codex via Databricks is managed through the Databricks Unity AI Gateway. This unified layer offers centralized control over permissions, rate limits, and cost management.

Unified Governance for Frontier AI

The Unity AI Gateway provides essential enterprise-grade controls. Features include granular permissions, configurable guardrails for content safety and PII detection, and comprehensive observability of all model interactions. This ensures that even sophisticated AI models can be deployed responsibly.

Databricks co-founder Patrick Wendell highlighted the partnership's value for enterprises seeking to leverage cutting-edge AI. OpenAI's CRO Denise Dresser also commented on the collaboration.

The Gateway also supports automatic failover and provides a single, consolidated bill for all model usage and coding tool interactions. It aims to simplify the management of complex AI deployments.

Putting GPT-5.5 to Work

Enterprises can now use GPT-5.5 for a range of demanding tasks. This includes complex document reasoning, long-horizon coding agent development, and enhancing data pipelines with AI-driven intelligence. The model's capabilities are particularly suited for agentic enterprise work.

Databricks' own natural language analytics experience, Genie, is powered by GPT-5.5 and other frontier LLMs. This allows business users to query complex enterprise data using plain English.

Developers can also build custom agents using Agent Bricks, leveraging GPT-5.5's advanced reasoning for multi-step workflows. This extends to automating departmental business processes and document analysis pipelines.

The integration of GPT-5.5 into Databricks' Lakeflow Spark Declarative Pipelines enables AI-powered ETL for document ingestion and transformation. This enhances reliability for complex data formats like scanned PDFs.

Databricks is making GPT-5.5 accessible across AWS, Azure, and GCP. Users can explore the model in the AI Playground, deploy Codex through the Unity AI Gateway, or build custom agents with Agent Bricks.

This collaboration represents a significant step in bringing advanced AI models into governed enterprise environments, building on Databricks' ongoing efforts to manage the complexity of AI.