Skip to content
NLEN
Illustration: a light, glowing atom-shaped AI chip racing along a circuit-board track, next to a long scroll of code.

Gemini 4 Hydrogen: Google's fast model with a 1 million token context at a flat price

By Ivo Donker · — text: Claude (AI, Anthropic), checked against the sources

What is Gemini 4 Hydrogen and what does it cost? It is Google's new fast, low-cost model, announced on November 12, 2026. It has a context window of 1 million tokens and costs 1.50 dollars per million input tokens and 7.50 dollars per million output tokens, the same price across the whole window. Google positions it as the successor to Gemini 3.8 Flash, aimed at everyday coding and agent tasks.

We read Google's announcement, written by Koray Kavukcuoglu of Google DeepMind. The topic was also in the trending briefing of November 12 and on the radar of November 12.

What Google announced

The model ID is gemini-4-hydrogen. Hydrogen sits next to the heavier Gemini 4 Carbon and Gemini 4 Argon models. There are four thinking levels: low, medium, high (default) and a new xhigh level. The context window is 1 million input tokens, with up to 1 million output tokens. The announcement does not state a knowledge cutoff.

The price is flat

Google writes: "Gemini 4 Hydrogen is priced flat across its entire 1 million-token context window". That stands out, because other providers charge a higher rate above a certain context length. We covered the heavier model earlier in our article on Gemini 4 Carbon. Anyone comparing models on price per task can work with a single rate, even when sending a whole codebase in one call.

Benchmarks: Google's figures

All figures below come from Google itself. On the Artificial Analysis Intelligence Index, Google says Hydrogen scores 51 at the default high level, against 50 for Argon, and 54 at xhigh. On Terminal-Bench 4.1 Google reports 57 percent, against 18.2 percent for Gemini 3.8 Flash. Google also cites 58.8 percent on HLE-Diamond and says Hydrogen finishes tasks 2.5 times faster than 3.8 Flash. According to Google, partners reported 2 to 2.8 times faster runs and 12 to 23 percent fewer tokens.

Where it is available

Hydrogen is available through the Gemini API and Google AI Studio, on all tiers. In the Gemini app it is in the model picker; free users get it through automatic model selection when the model fits the prompt. In Google Antigravity it is only for paying Google AI Pro and Ultra subscribers. It is also rolling out in Google Search and in Gemini in Google Sheets.

What this means

For agent loops, where a model processes many tokens often, the price per million tokens weighs heavily. Hydrogen is cheaper than Carbon and sits in the same price range as other fast models. Background on why long context can be expensive is in the race for context. Whether Hydrogen reaches the promised speed in practice remains to be shown by third-party measurements.

What we did not check

We did not test the model ourselves. All benchmarks and speed figures were reported by Google. We did not separately read the Gemini API pricing page for this article; the price comes from the announcement.

Checked on November 12, 2026.

This article was written with AI (Claude, Anthropic) and checked against the sources below.

Sources