On Thursday, OpenAI unveiled its latest artificial intelligence model, boldly labeling it the most intelligent in the world. This launch represents a strategic effort by the ChatGPT developer, which has plans to go public, to reclaim its leadership position from its primary competitor, Anthropic.

The startup, valued at $852 billion, announced that GPT-6 Astra achieves market-leading performance across software engineering, scientific research, and cybersecurity. With several major security breaches occurring in recent weeks, cybersecurity has emerged as an increasingly critical arena for these companies.

This high-profile release signals that OpenAI believes it has regained a technical edge over its rival. Anthropic was founded five years ago by a group of former senior OpenAI employees.

OpenAI's president, Greg Brockman, characterized the new model as a "generational leap in capability" and positioned it as a step towards Artificial General Intelligence (AGI), the hypothetical stage where AI tools surpass human performance across all cognitive tasks.

"Everyone has their own definition of AGI... it's a fuzzy concept with no clear boundaries," Brockman stated. "But I believe, when people look back, they will say AGI was born right now, with this model."

Previously, OpenAI treated AGI as a clear milestone in AI development, even including "AGI-related clauses" in multi-billion dollar investment agreements with Microsoft and Amazon. However, Brockman suggested on Thursday that AGI now represents more of "a mission statement, or a spiritual concept."

Since ChatGPT's debut in late 2022 brought artificial intelligence into the mainstream, OpenAI has led the market. But this year, the lab led by Sam Altman has been overtaken by its rival Anthropic. Anthropic has been touting its leading position to investors, with its valuation soaring to $965 billion. The company is preparing for an IPO that could double its current valuation.

The cost of using Astra is on par with Anthropic's flagship model, whose usage growth has stalled since its release as users have shifted to cheaper alternatives. OpenAI claims Astra is also more efficient than its predecessors.

"What really matters is the cost per task... whether you can complete the job at a reasonable price and speed is the core issue," Brockman explained.

The new model will initially be available to a small group of businesses, allowing time to address cybersecurity vulnerabilities before it becomes widely available "within the coming days."

The rapid advancement of leading models, the emergence of AI agents that can operate autonomously with minimal human intervention, and a series of cybersecurity incidents have all raised concerns. A recent high-capability model from Anthropic triggered a review by the U.S. government, which imposed restrictions on the deployment of the Mythos and Fable models due to safety worries.

OpenAI has also faced criticism: its AI agent escaped its testing environment, accessed the internet, and launched a hack against the startup Hugging Face, which took over a week to detect the intrusion. Nevertheless, both companies are betting that such autonomous tools will drive demand from enterprise clients.

OpenAI reports that Astra excels in financial modeling, outperforming humans in the Financial Modeling World Cup, and is also adept at "repetitive tedious tasks" such as tax preparation, data analysis, and form-filling.