Carbon and Argon think for a long time. Hydrogen is for the work that has to be done now. Google announced it on 12 November and shipped it 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.
Fact
The facts: on 12 November 2026 Google announced Gemini 4 Hydrogen, a model for everyday coding and agentic tasks, available to everyone the same day. According to Google, Hydrogen surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level: 51 vs 50, reaching 54 at its highest xhigh thinking 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. The model pairs native 1 million-token input with up to 1 million output tokens. According to Google, Hydrogen ranks #5 on Agent Arena, leads its class on HLE-Diamond (58.8%), ranks #1 overall on Vals Tax Agent Bench (52.3%) and ties for #1 on the Vals Legal Research Bench (55.8%). It scores 53.4% on the new MKA benchmark and 71.6% on Agents' Last Exam v1.1. Against Gemini 3.8 Flash, Hydrogen finishes tasks 2.5 times faster at its high default setting and more than triples Terminal-Bench 4.1 scores (57% vs 18.2%), cutting fabricated claims by more than 5x. It runs standalone across four configurable thinking levels or alongside Gemini 4 Carbon and Gemini 4 Argon.
The price is flat across the entire 1 million-token window. No tiers and no long-context discount: $1.50 per million input tokens and $7.50 per million output tokens, according to Google. For a model that takes in and returns up to a million tokens, that is a rare flat rate.
The new part of the announcement is not only the model. Google is introducing Orchestrator in Google Antigravity: instead of one model from start to finish, it switches between models like Carbon and Hydrogen inside the same session. According to Google, on Terminal-Bench 4.0 Orchestrator reaches near-Carbon task success rates while consuming 24% less quota on average and finishing 19% faster.
One model thinks, another executes. Google is selling the second.
Google's partners talk about the model in the announcement. Box says that in its evaluations Hydrogen ran about twice as fast as Gemini 3.8 Flash and used 12% fewer tokens. Those are their words, not ours.
The numbers are Google's. There is no independent check on the page. If you write code every day, the truth about Hydrogen will come out on your task, not on their table.
Official primary sources
Google - Introducing Gemini 4 Hydrogen for Agentic Workflows, 12.11.2026Google - Gemini 4 Hydrogen: everyday coding and agentic workflows for developers, 12.11.2026The visual is generated code-art, no third-party images.