Google announces Gemini 4 Argon with limited initial access

Google has announced Gemini 4 Argon, its latest AI model, but is initially limiting access to a “set of trusted cyber defenders.”

Google DeepMind chief AI architect and senior vice president Koray Kavukcuoglu said the model delivers “frontier performance” in complex workflows, including software engineering, enterprise knowledge work such as legal and finance tasks, and cybersecurity defense. He said the model is already powering Google’s internal workflows and has helped with large-scale codebase migrations.

Google also published a chart comparing Gemini 4 Argon with models from OpenAI and Anthropic across multiple benchmarks. The Verge noted that the AI community had been discussing what appeared to be leaked benchmarks earlier in the week.

Kavukcuoglu said Google is participating in the U.S. government’s voluntary process for pre-release model access as it gradually expands availability. Before a broader rollout, the company plans to strengthen safeguards against misuse and prompt injection attacks and to monitor for misalignment.

The announcement came one day after OpenAI’s DevDay, where the company announced its Dots AI agent and GPT-6.1 Sol AI model. OpenAI also said it would not release a planned GPT-6.1 Astra model, citing safety concerns. Kavukcuoglu was appointed DeepMind’s new head in August.