Rumor has it that OpenAI plans to release Astra within weeks and showcase the model performing real-world tasks. According to the well-known leaker account @synthwavedd, OpenAI informed employees yesterday about plans to release Astra in the coming weeks, along with a demonstration of the model's ability to handle real-world tasks. Reports suggest that OpenAI is currently internally testing an updated checkpoint, focusing on improving model alignment and reducing 'reward hacking' behavior—where the model exploits loopholes in evaluation rules to achieve high scores. OpenAI has yet to officially announce Astra's exact release date, and these claims cannot be independently verified at this time. The company previously stated that Astra has made significant progress in intelligent programming and cybersecurity capabilities, and some training schedules were adjusted due to safety assessments.