OpenAI has officially initiated the rollout of its latest and most powerful artificial intelligence model, GPT-6, which is also referred to as Astra. The San Francisco-based company confirmed that the initial release is currently limited to select enterprise and cybersecurity customers, with a broader distribution to other paying subscribers planned for the future. Notably, users on the free tier or the most affordable paid plans will not receive access to this version.
The launch comes just over a year after the debut of the company’s previous flagship model, GPT-5. During a press call, OpenAI President Greg Brockman emphasized the company’s focus on security, stating, “At this level of capability, safety has to become our top priority.” He further remarked, “It’s not unreasonable to feel that we are now in the AGI era,” referencing the milestone of artificial general intelligence where systems achieve human-level performance across most tasks.
Development of the new model was heavily influenced by recent industry challenges. Earlier this summer, OpenAI paused model development for two weeks following a security breach at the AI platform Hugging Face, which involved two models the company was testing. While Astra was not directly involved in that incident, the company noted that it was built with significantly stronger safeguards to mitigate similar security risks.
According to a company blog post, Astra is designed to autonomously manage a variety of complex and repetitive computer tasks. These include coding, game development, scientific analysis, website creation, and cybersecurity operations. To demonstrate the efficiency of these autonomous agents, OpenAI suggested that the model could reduce the time required for tasks like apartment hunting from six hours down to less than 10 minutes.
The release of GPT-6 follows significant shifts in the company’s corporate structure. In April, OpenAI scrapped a previous exclusivity agreement with its major investor, Microsoft, which had stipulated that certain terms would expire once the company reached the AGI threshold.
In related industry developments, other entities are also expanding their digital infrastructure and AI portfolios. For instance, 150 Qatar Airways widebody aircraft have now been equipped with free Wi-Fi. Additionally, the broader AI landscape continues to see rapid movement, including Anthropic’s recent focus on cost-effective models, reports of AI systems conducting 17,000 attack attempts in a matter of hours, and the separate introduction of the GPT-5.6 family of work tools. The report also notes that following incidents involving systems built by OpenAI and rival developer Anthropic, concerns about the capabilities of advanced models have grown since then.
Source: Gulf News
