To meet the growing organizational demand for agentic AI, OpenAI is releasing a new series of open-weight models designed for executing complex tasks. These models provide a strong performance-to-size ratio and feature advanced reasoning capabilities, including adjustable logic levels and chain-of-thought outputs that break down problems into manageable steps. This makes them ideal for use in agentic workflows, coding, scientific analysis, and mathematical problem-solving.
The models are equipped to follow instructions and use tools like web search and a code interpreter, enabling them to reference real-time information and perform multi-step operations. With a 128K context window, they can also process lengthy documents and conversations, such as customer service transcripts, detailed technical documentation, and academic papers. OpenAI has emphasized safety as a core component, noting that each model has undergone comprehensive training and evaluation to support responsible deployment.
Source link
