OpenAI’s September 1 blog post confirming that Astra, its next major frontier model, has reached the “Critical” cybersecurity capability threshold under the Preparedness Framework is the dominant driver behind the market’s heavy weighting toward an imminent release. The company stated it has completed additional safety testing and will make Astra “available soon,” with advanced cyber capabilities initially limited to select Daybreak Blue testers and a full system card published at launch. This follows weeks of paused development after internal evaluations flagged autonomous exploit-generation risks, pushing implied probabilities sharply earlier in recent days. Traders interpret the phrasing and timing—coupled with OpenAI’s pattern of rapid iteration on frontier models—as evidence of a launch window in the next 24–48 hours, while the low odds on “no release by September 30” reflect the absence of further announced delays.