Databricks brings GPT-5.5 to enterprise agent workflows

Source: OpenAI Date: 2026-05-15 URL: https://openai.com/index/databricks

Summary

OpenAI and Databricks announce that GPT-5.5 is being integrated into Databricks’ enterprise data and AI platform, making it available to the companies already running data pipelines, feature stores, and ML workflows there. The source page returned 403; the signal is carried by the title and the broader pattern of OpenAI’s enterprise partnership announcements following the GPT-5.5 launch. The partnership positions GPT-5.5 as an agent-ready model available where enterprise data already lives, rather than requiring data to move to an OpenAI-managed surface.

Implications

This feeds the enterprise AI platform consolidation thread and is the clearest signal yet of OpenAI’s strategy for the data-platform layer.

• Meet the data where it lives. Databricks is where a significant portion of enterprise ML infrastructure runs. Embedding GPT-5.5 there means OpenAI is competing with a “bring the model to the data” motion rather than a “bring the data to the model” one — which is the same move Anthropic is making via AWS Bedrock and the Claude for Work integrations.
• Agent workflows as the hook. The framing is “agent workflows,” not “completion API” — this is a bet that Databricks customers will run autonomous agents on top of their existing pipelines, and that GPT-5.5’s reasoning depth is what enables that. The partnership is a distribution play dressed as a capability play.
• Watch: Whether this accelerates similar partnerships — Google (Vertex AI), Snowflake, and MongoDB are all plausible next announcements. If every major data platform announces a flagship model partnership in 2026, model selection at the enterprise level effectively becomes a data-platform vendor decision.