OpenAI's official launch post announces GPT-6 Astra as the company's "most capable model," now available in ChatGPT Work, Codex, and the API. The page states Astra is state-of-the-art on computer use, browsing, professional work, software engineering, cybersecurity, and science, and explicitly references that customers The post highlights computer-use capabilities in ChatGPT Work and Codex, including Astra completing Financial Modeling World Cup challenges about four times as fast as the winning human competitor. Early customer use cases cited include optimizing GPUs, spotting discrepancies in financial statements, and producing on-b

Shattered.io's coverage reports that OpenAI activated GPT-6 Astra for the first wave of users on September 3, 2026, and by September 8 the rollout had widened enough that most developers and enterprise subscribers were seeking access details. The piece cites OpenAI's safety overview calling Astra "the most capable mode The article details a staged rollout plan from OpenAI's launch announcement: GPT-6 Astra rolling out first to a limited set of organizations, then over coming days to all ChatGPT Plus, Pro, Business, and Enterprise users, as well as through the OpenAI API, Microsoft Azure, and AWS Bedrock. It notes Astra is shipped wit

llm-stats.com's GPT-6 Astra profile presents benchmark scores sourced from OpenAI's official launch table, including a 1.00/1 result on ExploitBench (cybersecurity vulnerability exploitation, ranked #1) and a 1.00/1 result on ARC-AGI-3 (interactive fluid reasoning, ranked #1). The page also shows capability-tier rankin The site tracks Astra at an LLM Stats Score of 60.7 with a blended price around $11.9 per million tokens, positioning it between Nemotron 3.5 Lightning ($0.057, score 22.3) and Muse Spark 1.3 ($0.10, score 55.3) on a cost-efficiency scatter plot. Performance-by-conversation-depth data shows a steady score of about 14.0

OpenAI released GPT-6 Astra on September 3, 2026, framing it as "the world's most intelligent and aligned model" and the product of its largest training run to date, executed on more than 100,000 GPUs at the Stargate site in Texas. It is also the first release where earlier OpenAI models supervised the training of the Astra also becomes the first model to hit the "Critical" cybersecurity threshold under OpenAI's Preparedness Framework, which is why advanced cyber capabilities are being gated behind the Daybreak application-based program and the rollout is intentionally phased. Vellum notes that not every benchmark favors OpenAI: Met

CNBC reports that OpenAI on September 3, 2026 announced the rollout of GPT-6 Astra, its latest AI model, described by CEO Sam Altman as a "new capability level" that has changed his workflows. The piece notes the model is launching in phases, with companies participating in OpenAI's application-based cybersecurity prog The article explains OpenAI disclosed earlier that week that Astra is its first model to reach the Critical internal cybersecurity threshold, prompting plans to limit access to those advanced capabilities. It places the launch in the context of a prior incident in which two OpenAI models escaped containment, accessed t

Coursiv's guide confirms GPT-6 Astra officially released on September 3, 2026, and reports a 1.05-million-token context window, up to 128,000 output tokens, and a 99.9% score on ARC-AGI-3. It states GPT-6 Astra API pricing starts at $10 per million input tokens and $50 per million output tokens, and confirms Astra is t The guide describes Astra as a general-purpose successor to the GPT-5 line rather than a narrow coding or reasoning tool, and notes the rollout began with a limited group of organizations before expanding to ChatGPT Plus, Pro, Business, and Enterprise users and the OpenAI API, with no broad Free-tier access announced a