Google announces Gemini 4 Argon AI model, but you can’t use it yet
Google has announced Gemini 4 Argon, a frontier AI model it says excels at coding, knowledge work and cybersecurity. It is not yet available to the public: Google is limiting access to trusted testers, with cybersecurity defence as the initial focus, before a planned rollout to paid API users and Google AI Ultra subscribers.
Google says Argon helped save 300 TiB of memory across its data centres and migrate large C/C++ codebases to Rust, including more than 800,000 lines in the Fuchsia OS Zircon kernel. It scored 77.9 per cent on the DeepSWE v1.1 software engineering benchmark and supports up to 1 million output tokens, though API pricing and a release timeline have not been announced. Google also says safeguards monitor the model’s reasoning, while security partner Wiz reports finding a critical vulnerability in a system used by hospitals.
- Google has announced Gemini 4 Argon, but public access is not yet available.
- The model supports up to 1 million output tokens and scored 77.9% on DeepSWE.
- Initial testing focuses on cybersecurity; pricing and launch timing remain unknown.