Top News

Tech company OpenAI has unveiled its most advanced and powerful flagship model to date, 'GPT-6 Astra'. The company claims that this new model represents a major milestone on the path toward Artificial General Intelligence (AGI). Announcing the launch on the social media platform X, OpenAI CEO Sam Altman hailed it as the world's best model for professional tasks, scientific work, computer usage, coding, and cybersecurity.

During a media briefing, OpenAI President Greg Brockman stated that it is entirely accurate to say we have entered the era of AGI. He believes that future generations will look back on the launch of Astra as a pivotal moment in the history of AI. Sam Altman echoed this sentiment, noting that the model would accelerate the growth of a new generation of startups, scientific discoveries, and technological innovations at an unprecedented pace.

**Surpassing Anthropic's Model in Coding**

OpenAI has described GPT-6 Astra as the best software engineering model ever created. According to benchmark results released by the company, Astra has significantly outperformed its predecessor, GPT-5.6 Sol, as well as Anthropic's Fable models, in complex tasks such as bug detection and terminal execution. The model achieved impressive scores, including 98 percent in Frontier Math Tier-4 and approximately 99.9 percent in ARC-AGI 3.

**Capable of Handling Research, Website Design, and Tax Returns**

Astra's capabilities extend far beyond coding. The model can perform a wide range of tasks with complete autonomy based on user instructions. It excels at conducting online research, analyzing scientific data, filling out online forms, and managing calendars. Furthermore, it can build entire websites from scratch, install and test software, draft legal documents, and even prepare tax returns. In a video released during the launch, Astra successfully performed all these tasks without any human assistance.

**Cybersecurity and Robust Safety Features**
Given the AI-related cybersecurity threats in the tech world, the company has implemented stringent security measures in Astra. In tests conducted without production safeguards, it achieved a 100 percent score on 'ExploitBench'. However, following a previous incident involving Hugging Face, OpenAI has paid special attention to its safety alignment. The company claims that this is its safest and most controlled model to date.

OpenAI has begun rolling out GPT-6 Astra to customers of its 'Daybreak' cybersecurity program. Within the next week, the model will also become available to ChatGPT Plus, Pro, Business, and Enterprise subscribers. Additionally, developers will be able to access it via the OpenAI API, AWS, and Amazon Bedrock. Regarding the phased release, Sam Altman apologized for the delay and assured that it would soon be made fully available to everyone.

Disclaimer: This content has been sourced and edited from Amar Ujala. While we have made modifications for clarity and presentation, the original content belongs to its respective authors and website. We do not claim ownership of the content.