OpenAI has introduced GPT-6 Astra, describing it as the world’s most intelligent and aligned model to date. The model is the result of OpenAI’s years of research and investments in pretraining, reinforcement learning, and alignment. GPT 5.6 Astra stands as a major step in AI reasoning, software engineering, computer use, cybersecurity, scientific research, and professional work.
It can carry out tasks such as filling out online forms, updating customer records in CRM, conducting research, testing software, and producing documents, spreadsheets, and presentations.
Record Performance Across Benchmarks
Astra delivers state-of-the-art performance across several benchmarks. The intelligent model scored 98% on FrontierMath Tier 4 and saturated ARC-AGI-3 with a 99.9% score and ExploitBench with a 100% score. It also showed improvements in real knowledge work tasks.
In latency simulations on OSWorld 2.0 Astra achieved higher computer-use performance in about 47% less time per task, scoring 72.6% in around 40 minutes per task as compared with 65.7% in 75 minutes for GPT-5.6 Sol. One of the standout features is GPT-6 Astra’s computer and browser use, which handles speed, accuracy, and safety effectively.
A Major Focus on Alignment and Safety
OpenAI says Astra is the most aligned model. It excels at exercising care, respecting task boundaries, and communicating transparently. In one of the evaluations, it was found that as compared to GPT-5.6 Sol, which without production safeguards went beyond the authorized target of 48% of the time, GPT-6 Astra did so in 0% of cases. As an additional layer, the company has built system safeguards such as Codex Auto review and monitoring agents' reasoning and actions to detect and contain unsafe behavior.
Cybersecurity Capabilities and Safeguards
Another point is that Astra is a major upgrade in cyber capabilities. OpenAI says the model meets the Critical threshold in cybersecurity under its Preparedness Framework. Astra possesses the ability to identify and develop zero-day exploits to help defenders find and patch weaknesses; however, also creates a need for stronger safety.
To address these risks, OpenAI says the model will decline advanced cybersecurity tasks such as creating proof-of-concept exploits for vulnerabilities, while supporting defensive workflows, like malware analysis, proof-of-concept validation, and detection engineering.
Availability and Pricing
GPT-6 Astra is initially rolling out to a limited set of organizations and will be available to all ChatGPT Plus, Pro, Business, and Enterprise users in the coming days. It will also be available through the OpenAI API, Microsoft Azure, and AWS Bedrock. Users on the Pro, Business, and Enterprise plans will have access to Astra Pro. For developers, the model is available as gpt-6-astra, and through Microsoft Azure and Amazon Bedrock.
OpenAI Standard pricing ranges from $10 per million input tokens to $50 per million output tokens. Overall, GPT-6 Astra marks a major shift toward AI systems that can do more than answer questions and get real work done with all its advanced capabilities.
Read more trending news updates on our website. We help you navigate through complex tech with ease.
Recommended For You:
OpenAI Introduces GPT Image 1.5: A Sophisticated Image Generation and Editing Model
