Google's parent company, Alphabet, has launched Gemini 4 Argon. This is a new AI model designed to handle a variety of tasks such as programming, research, and writing. However, Google states that Argon is particularly adept at cybersecurity.

Google stated that Argon is currently only available to a few cybersecurity partners within the company Fairwind Program. Fairwind Program is Google's security initiative. This model is specifically trained for defensive cybersecurity tasks, and the company claims that it can "autonomously discover, verify, and fix critical software vulnerabilities."

Google also stated that Argon performs exceptionally well in programming and engineering, and noted that company employees are already using this model in their daily work, including for debugging and code library migrations. Google also emphasized the ability of Argon to parse visual content – whether it's analyzing long videos or charts.

In a blog post on Wednesday, Google stated: “Argon has the capability to perform deep reasoning continuously in complex, long-duration workflows, and is fundamentally changing the way we work and build things at Google.”

Top-tier AI laboratories are competing to release increasingly powerful models, striving to outperform each other, although these companies also warn that AI could get out of control. Not long ago, OpenAI released Astra, claiming it to be the best model to date. Earlier this year, Anthropic also released Fable, using similar promotional rhetoric.

These companies are all striving to promote their services as being superior to those of their competitors. In their blog, Google claims that Argon scored significantly higher than GPT-6 of OpenAI, as well as Fable and Opus models of Anthropic in multiple AI benchmark tests. Google also cited data from a AI benchmark test startup, Vals, stating that Argon currently holds a leading position in the company's AI model index.

Google was once considered to be "lagging behind" in the " AI competition," but recently Gemini has achieved success. In August, the company announced that it had over 1 billion monthly active users, a metric that allows it to compete with OpenAI; the latter also just announced that its monthly active users had reached 1 billion for ChatGPT.