Google unveils its most powerful AI model, Gemini 4 Argon — CNBC

Alphabet has unveiled Gemini 4 Argon, a new artificial intelligence model that it called its most powerful development. It is initially planned to be made available to trusted cybersecurity partners, CNBC Top News reports.

Model capabilities

Google said Gemini 4 Argon significantly improves capabilities in programming, cybersecurity, and completing complex professional tasks. According to the company, the model set a new record in real-world software development tasks and also shared first place in cybersecurity evaluations.

The company also claims that Argon achieved the best result in another test measuring performance in financial, legal, and other professional tasks. Google said the model outperforms the previously released 3.8 Flash Cyber in identifying vulnerabilities.

More current news is available on the UA.News Telegram channel Telegram.

Use and security

Gemini 4 Argon is already being used inside Google to optimize memory usage in data centers. According to the company, this made it possible to free up hundreds of terabytes of memory without purchasing additional hardware. The model was also used by quantum computing researchers.

Google plans to deploy Argon in stages, starting with trusted cybersecurity partners. The company is conducting a pre-release security assessment with the U.S. government and said it intends to strengthen protections in four areas, including against misuse and prompt injections, before the model’s public launch.

According to CNBC, the release came the day after Google CEO Sundar Pichai signed a voluntary agreement on AI safety with U.S. President Donald Trump. Before that, the White House hosted a meeting with leaders of major technology companies devoted to artificial intelligence safety issues.