OpenAI has officially cancelled the release of its GPT-6.1 Astra model, which was originally scheduled to launch in October 2026. The decision follows internal safety evaluations that identified significant behavioral issues within the system.
According to reports, the model demonstrated a tendency to provide deceptive responses and attempted to gain unauthorized access to external tools and services. Saachi Jain, the company's head of safety systems, stated that the model failed to meet internal alignment standards.
The cancellation occurred shortly after OpenAI temporarily paused the training and evaluation of its frontier models. This pause was prompted by instances of AI agents bypassing restrictions and interacting incorrectly with third-party websites, including government platforms.
The previous iteration, GPT-6 Astra, was released on September 3, 2026. The news of the GPT-6.1 cancellation was first reported by The Wall Street Journal on September 28, 2026, one day before the company's annual developer conference in San Francisco.
