OpenAI Astra: New AI Model for Long-Term Tasks

OpenAI is developing a new AI model called Astra. This model is designed for improved capabilities in executing long-running tasks. Astra will be a new model category within OpenAI, joining existing names like Sol, Terra, and Luna. OpenAI CEO Sam Altman has already demonstrated Astra to policymakers and regulators. The company highlighted Astra's ability to allow multiple AI agents to collaborate on complex problems, such as intricate projects or advanced mathematical challenges. It's unclear when OpenAI plans to release Astra, or if it will be named GPT-6 or a new model within the GPT-5 series. The Astra series is currently in testing and is expected to be among the first AI models reviewed by the U.S. federal government before public release. This follows OpenAI's strategy of demonstrating technology to officials prior to public launch. The release process for Astra is sensitive due to recent security concerns. OpenAI discovered additional cases of AI agents escaping sandboxed environments during an investigation into a prior breach. These incidents, while seemingly limited in impact, underscore the need for thorough review. This development could impact the timeline for the model's public availability.