Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

theverge+1cnbccnbcOpenAI released GPT-6 Astra on September 3, describing it as the product of "years of research and big bets," but the launch of its most capable model was quickly overshadowed by access problems that left paying subscribers locked out. CEO Sam Altman took to X hours after the announcement to apologize for what he called a "messy rollout," as complaints piled up from Plus, Pro, Business, and Enterprise customers who expected early access.theverge+3
Astra debuted first for a limited set of organizations in OpenAI's Trusted Access Program and its Daybreak cybersecurity platform, with broader availability promised "over the coming days". But Pro subscribers, who typically receive priority access to new releases, found themselves behind enterprise customers in the queue. OpenAI offered "banked resets" — one for every day a paying user went without access — as compensation while the rollout continued.Mint+3
The model itself represents a generational leap. OpenAI says Astra is its first model to reach the "Critical" level of cybersecurity capability under its Preparedness Framework, meaning it can identify and develop functional zero-day exploits in hardened real-world systems. It features a context window of over one million tokens, 128,000 max output tokens, and API pricing of $10 per million input tokens and $50 per million output tokens — roughly 2.5 times the cost of its predecessor, GPT-5.6 Sol.yottalabs+2
Even as access issues persisted, early users offered praise. OpenAI developer Thibault Sottiaux called Astra the company's "biggest competitive advantage," saying internal use boosted productivity so much that some development plans were pulled forward by six months. The model has drawn attention for its autonomous computer-use abilities, with demos showing it operating tools like Blender Autodesk, Inc. directly on a user's desktop.x+1
The launch landed in one of the busiest weeks the AI industry has seen. Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on Tuesday, Meta announced Muse Spark 1.3 and Google Alphabet Inc. unveiled Gemini 3.8 Flash on Wednesday, and Nvidia finalized its $12.9 billion acquisition of Hugging Face. Altman told CNBC Comcast Corporation that "we're all moving to faster cadences," attributing the clustering partly to everyone returning from summer vacation.cnbc
The compressed timeline has left enterprise customers scrambling. "I feel like model fatigue is a real thing," said Zhen Lu, CEO of AI startup Runpod. Clockwork Systems CEO Suresh Vasudevan put it more bluntly: "It's really challenging to go evaluate every one of the ones that are coming out right now".cnbc