Google launches Gemini 4 Argon, initially restricted to trusted cyber defenders
Google is staging access to its new flagship model on safety grounds; all performance claims are self-reported, so both capability and rollout pace remain to be seen.
Google released its next flagship model, Gemini 4 Argon, on September 30, initially limiting access to a set of “trusted cyber defenders,” which the company says is to check for misalignment before broader rollout.
Koray Kavukcuoglu, chief AI architect at Google DeepMind, said the model delivers frontier performance in software engineering, enterprise knowledge work such as legal and finance, and cybersecurity defense, and that it already powers internal workflows like large-scale codebase migrations. Google also said it is engaged in the U.S. government's voluntary process for pre-release model access.
The benchmarks Google shared show it beating competing models from OpenAI and Anthropic, but those figures are self-reported and not independently verified, and no timeline for wider availability was given.
Sources:https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argonhttps://www.theverge.com/tech/1002980/google-gemini-4-argon