Google announces Gemini 4 Argon AI model, but you can't use it yet

Google has unveiled Gemini 4 Argon, its latest AI model, though it remains unavailable to external users. The company claims Argon outperforms competitors in programming, knowledge work, and cybersecurity, based on a series of internal tests. Google engineers are already using the model to fine‑tune their own systems; for instance, Argon reportedly processed fleet‑level telemetry data, saving roughly 300 TB of memory in the firm’s data centers. The model has also helped migrate C/C++ code to Rust across Google’s core libraries and the Zircon kernel of the Fuchsia operating system, converting more than 800,000 lines of code.

In published benchmarks, Gemini 4 Argon scored 77.9 % on the DeepSWE v1.1 software engineering test, outperforming other recent models such as GPT‑6 Astra, Fable 5.1, and Opus 5.5. It also achieved the highest score on the Vals economic analysis index, supporting Google’s claim of strong long‑term task performance.

This development is significant because it signals rapid progress in building more powerful and efficient AI models, which could translate into substantial gains in developer productivity and the security of critical infrastructure, even as access remains restricted during ongoing internal testing.

Read the original article on Ars Technica AI

This summary is an informational synthesis produced by dataqbs.com. All rights to the original content belong to its author and the cited media outlet. We act solely as curators of technology news and claim no authorship.