Google unveils Gemini 4 Argon
Gemini 4 Argon leads on benchmarks and powers coding, charts, video, and long docs See why Google is limiting access now and what its API pricing signals for teams
Google has introduced Gemini 4 Argon, a new frontier AI model that the company says performs strongly on its internal benchmark tests. The model is rolling out first to a limited set of vetted cybersecurity teams rather than to the general public.
According to the article, Argon ranked highly on several leaderboards and performed well on tasks involving coding, long documents, charts, and video. Google has not announced a broader launch date, although it has shared API pricing for the model.
The newsletter also notes some internal skepticism about whether the model’s strong benchmark results will translate into realworld performance. Google rejected that criticism, but the article says the broader rollout will be important in assessing the model’s impact.