Databricks partners with OpenAI on GPT-5.5

Databricks partners with OpenAI to introduce GPT-5.5, a frontier model for enterprise agentic work and complex reasoning.

• •GPT-5.5 powers Codex coding agent with enhanced reasoning and execution capabilities for developer workflows
• •Handles multi-step tasks including code writing, debugging, research, data analysis, document creation, and tool operations
• •On OfficeQA benchmark, GPT-5.5 achieved 64.66% accuracy, 13% improvement over GPT-5.4's 57.14%
• •In full-agent workflows, GPT-5.5 scored 52.63% with 46% error reduction compared to GPT-5.4's 36.10%

This summary was automatically generated by AI based on the original article and may not be fully accurate.