Source: OpenAI's Astra model uses "recurrent depth", a technique that improves cost and performance but obscures the AI's reasoning, making it harder to monitor (The Information)

OpenAI’s upcoming Astra model employs a technique called “recurrent depth,” which reduces computational cost and boosts performance while making the model’s internal reasoning less transparent and harder to audit. The architecture also enhances capabilities in code generation and the ability to run applications on a computer, representing a functional upgrade over previous models.