Databricks announced native support for OpenAI’s GPT-5.5 on its platform, available now across AWS, Azure, and GCP, according to a post from Databricks co-authors Patrick Wendell, Hanlin Tang, Ahmed Bilal, Arnav Singhvi, Ivan Zhou, and Harish Gaur.
All GPT-5.5 usage on Databricks, including inference and Codex coding workflows, is routed through Unity AI Gateway, which Databricks describes as providing centralised security, cost controls, and observability across agents, queries, and coding workflows.
What Databricks customers can do with GPT-5.5
The post outlines four use cases Databricks is positioning for the integration.
Genie natural-language analytics. Genie, Databricks’ natural language analytics experience, is described as powered by multiple frontier LLMs including GPT-5.5 and other providers. Business users interact with enterprise data in plain English to explore data, answer ad hoc questions, and automate knowledge work. Databricks states that Genie’s understanding of enterprise data ontology, combined with GPT-5.5, allows business users to bring data insights into everyday work.
Agent Bricks custom agents. With GPT-5.5, Agent Bricks Custom Agents can handle more complex multi-step workflows, according to the post — from document analysis pipelines to automating departmental business processes. Developers can build GPT-5.5-powered agents using their preferred tools and frameworks, then deploy them as serverless Databricks Apps.
Document intelligence pipelines. The post describes chaining GPT-5.5 into Lakeflow Spark Declarative Pipelines for what Databricks calls “GenAI ETL” — ingesting documents, applying AI transformations such as summarisation, extraction, or classification, and orchestrating the flow with built-in governance and observability. Databricks states that GPT-5.5’s document parsing and grounded reasoning make these pipelines more reliable for complex artifacts including scanned PDFs, tables, and multi-format data.
Codex coding workflows. Databricks states that Unity AI Gateway governs both GPT-5.5 model inference and Codex coding workflows through a single control plane. The post links to separate documentation on how Unity AI Gateway governs agentic coding.
Governance framing
Databricks positions Unity AI Gateway as the enterprise control layer differentiating its GPT-5.5 offering from direct OpenAI API access. The post does not detail the specific security, cost control, or observability mechanisms that Unity AI Gateway applies in this configuration.
The announcement does not specify pricing for GPT-5.5 access through Databricks, token throughput commitments, or the relationship between Databricks compute costs and underlying OpenAI API charges.
Databricks co-founder and VP of Engineering Patrick Wendell is cited alongside OpenAI CRO Denise Dresser in the post as discussing “how the OpenAI-Databricks partnership creates unique value for enterprise customers,” though no quotes from either are included in the available excerpt.