OpenAI has introduced GPT-6 Astra, describing it as its most capable artificial intelligence model to date, with major advances in reasoning, software development, computer use, research and cybersecurity.
The company said GPT-6 Astra represents a significant step forward in AI capabilities and is designed to handle complex, multistep tasks that traditionally require human involvement.
It can work across areas including coding, scientific research, professional workflows and autonomous computer use.
OpenAI said Astra can identify previously unknown security vulnerabilities and develop exploits in certain circumstances.
The model achieved a 100% score on the ExploitBench benchmark, although the company cautioned that the result may be affected by possible exposure to vulnerabilities included in the benchmark.
The company has introduced additional safeguards alongside the model’s release because of its increased cybersecurity capabilities. OpenAI said it strengthened protections against harmful cyber activity and added measures including stricter isolation, enhanced monitoring and additional alignment evaluations.
GPT-6 Astra can also interact with computers, browse the web, work with documents and create spreadsheets and presentations. OpenAI said the model can follow templates and instructions while adapting when requirements change.
OpenAI said Astra is more difficult to misuse or jailbreak than its predecessor, GPT-5.6 Sol, and described it as its most aligned model.
In one internal evaluation, Astra did not go beyond an authorized task when faced with difficult or impossible instructions, compared with a substantially higher rate for GPT-5.6 Sol without production safeguards.
GPT-6 Astra began rolling out Sept. 3 to a limited group of organizations. OpenAI said broader access for ChatGPT Plus, Pro, Business and Enterprise users, as well as through its API and cloud platforms, will follow in the coming days.