we_are_coded.by CODE · The world, decoded
БГ

Concept

Gemini 4

Google's model family for deep thinking and fast execution. Carbon and Argon think long, Hydrogen executes.

Updated on 13 November 2026 · Checked on 13 November 2026

Gemini 4 is Google's family of models for agentic work. Three models with different roles: Carbon and Argon think long on the hardest problems, Hydrogen is for fast execution - writing code, running tests, checking changes.

Hydrogen launched on November 12, 2026 and is available to everyone the same day: through the Gemini API in Google AI Studio and Android Studio, in Gemini for business, in the Gemini app, in Search, and for paid Google AI Pro and Ultra subscribers in Google Antigravity. According to Google, it surpasses Argon on the Artificial Analysis Intelligence Index at its default thinking level: 51 vs 50, reaching 54 at the highest xhigh setting.

The price is flat across the entire 1 million-token context window: $1.50 per million input tokens and $7.50 per million output tokens. No tiers and no long-context discount. The model accepts up to 1 million input and up to 1 million output tokens.

Hydrogen runs standalone across four configurable thinking levels or alongside Carbon and Argon. Google also launched Orchestrator in Google Antigravity: instead of one model from start to finish, it switches between models like Carbon and Hydrogen in the same session. According to Google, on Terminal-Bench 4.0 Orchestrator reaches near-Carbon success rates while using on average 24% less quota and finishing 19% faster.

The numbers are Google's and there is no independent verification on the page. Partners also speak in the announcement: Box says that in their evaluations Hydrogen is about twice as fast as Gemini 3.8 Flash and uses 12% fewer tokens. Those are their words, not ours.

Primary sources

Google's announcement

The visual is generated code-art, no third-party images.