Google says Gemini 4 Argon helped Chrome engineers do months of work in weeks
Google DeepMind says its latest model helped Chrome engineers finish a months-long refactor in weeks.

- Engineers show it large blocks of old code
- It makes several clean changes to many files at once
- The whole project finishes in weeks instead of months
Google DeepMind has unveiled Gemini 4 Argon, the latest version of its Gemini large language model family, and it is already showing real-world impact far beyond chat applications. According to Google DeepMind, the model helped the Chrome engineering team at Google separate the code that handles webpages from a legacy folder of about 6 million lines, inside a Chromium codebase of 40 million lines 1. Making several clean code changes at a time, it helped the team finish a refactor across hundreds of files, doing months of work in weeks, the company says .
Google says the change will let live webpages move between tabs, side panels and floating windows without reloading, and remove a whole category of browser crashes 1. These are the company's own figures; it has not published a comparison with other models for this work.
Corrections and updates
- Corrected on 21 October 2026. An earlier version of this story included details its sources do not support. They have been corrected or removed.
- Corrected on 21 October 2026. The headline said Gemini 4 Argon beats rival models in intelligence tests, which the sources do not say, and a release date was given that they do not confirm. The headline now says what Google DeepMind reported, the details of the Chrome work are back with their source, and the date has been removed.
Key facts
- Gemini 4 Argon helped refactor a 6 million-line legacy Chrome folder, finishing in weeks what would normally take months
- Live webpages should move between tabs, side panels and floating windows without reloading, Google says
- Part of Google DeepMind's rapidly evolving Gemini model family
Sources
How fast the Chrome refactor finished, with and without the AI
Google DeepMind figures for the Chrome engineering team's refactor of Chromium; no external comparison published
If this story concerns you or your company, or uses something that belongs to you, ask for a correction or for it to be taken down. How requests work
Keep reading
Apps and productsGoogle opens TorchTPU, a native PyTorch backend for cloud TPUsHerald of AIhttps://github.com/soumith, CC BY-SA 4.0, via Wikimedia Commons
Google opens TorchTPU, a native PyTorch backend for cloud TPUs
A new open-source stack lets developers run PyTorch models directly on Google's TPU chips, replacing the older PyTorch/XLA system.
Business and rulesMirendil raises $1 billion to use AI agents for faster model developmentHerald of AI
Mirendil raises $1 billion to use AI agents for faster model development
A new startup founded by former Google DeepMind and Anthropic researchers says its automated platform lets small teams build large AI models quickly and cheaply.
New modelsGoogle's Gemini 4 Argon brings deep physical reasoning to real robotsHerald of AI
Google's Gemini 4 Argon brings deep physical reasoning to real robots
The robotics-focused Gemini 4 Argon model scores 87% on a hard 3D reasoning benchmark and controls Boston Dynamics Spot in real-world manipulation tasks.