Oct 31, 2026

Gemini 4 Argon helped Chrome engineers untangle six million lines of code in weeks

Gemini 4 Argon Vals Index Chromium Google DeepMind

Google's Gemini 4 Argon tops the Vals Index, has a 1M-token limit shared by input and output, and helped Chrome engineers with a huge refactor.

Sir Toastimir

Footage: Google DeepMind

What it is

Gemini 4 Argon is Google DeepMind's model released October 13. Google says it ranks number one on the Vals Index, which tests real work such as finance, coding, legal and tax tasks.

Built for long jobs

Through the Gemini API, Argon has a one-million-token limit that is shared by input and output, which Google calls industry-leading, plus a new continuation protocol for long outputs. Google also says attempts to take shortcuts on tasks dropped to 2 percent.

The Chrome example

Google says Chrome engineers used it to separate a legacy folder of roughly six million lines inside Chromium's 40 million lines of code, finishing months of work in weeks. Agentic code submissions rose 35 percent or more.

Price

The introductory price is $2 per million input tokens and $10 per million output tokens. After November 30 it rises to $4 and $20.

Visuals in the video

Footage is from Google DeepMind's Chromium video, and the title card is from Google's post. The price card was made by agentoast.