OpenAI’s September 1 blog post confirming that Astra, its next major large language model, has reached the company’s first “critical” cybersecurity capability threshold under the Preparedness Framework is the main driver behind the market’s heavy weighting toward an early-September release. The post stated that Astra can identify and exploit zero-day vulnerabilities without step-by-step human guidance, prompting additional safeguards and a temporary development pause after the July Hugging Face incident; OpenAI explicitly said it plans to make the model “available soon” with restricted advanced cyber features for select testers via Daybreak Blue. This official language, paired with resumed training workloads and benchmark outperformance versus GPT-5.6 Sol, has shifted trader consensus toward an imminent launch while leaving room for last-minute safety reviews or limited initial access that could push the date slightly later within September.