Alphabet has launched Gemini 4 Argon, a new artificial intelligence model that Google says is designed for coding, research and writing, with a particular focus on cybersecurity.
Argon is being rolled out only to a select group of Google cyber partners through the company’s Fairwind Program, its security initiative. Google said the model was trained specifically for defensive cyber work and can “autonomously find, validate, and patch critical software vulnerabilities.”
Google said Argon is also strong at coding and engineering. The company said its employees have already used the model in daily work, including debugging and codebase migrations.
Google also said the model can parse visuals, including the contents of long videos and charts. In a blog post published Wednesday, the company said Argon was “built to sustain deep reasoning across complex, long-horizon workflows.”
Google said Argon scored significantly higher than OpenAI’s GPT-6 Astra and Anthropic’s Fable and Opus models across multiple AI benchmarks. The company also cited Vals, an AI benchmarking startup, as showing that Argon currently leads on its AI model index.
OpenAI recently released Astra and described it as its best model yet. Earlier this year, Anthropic released Fable with similar claims about its performance.
Google, once seen as trailing in the AI race, has recently posted gains with Gemini. The company said in August that the Gemini app had more than 1 billion monthly users, a figure that puts it alongside OpenAI’s ChatGPT, which also recently reached 1 billion monthly users.
Featured image credit
