Google launches the low-cost model Gemini 4 Hydrogen, which scores 57% on the Terminal-Bench 4.1 terminal-task benchmark in Google's own test, against 18.2% for the previous Gemini 3.8 Flash

Concept illustration: a large black-framed board stands on two legs, and along a line across its lower part sit, from left to right, a black rounded block lettered FLASH, a short stack of a few cream sheets with grey lines on the top one, a second black block lettered HYDROGEN, and a much taller stack of red sheets whose top sheet also has grey lines. A small white sign at the board's lower right shows a multicoloured letter G next to the lettering GOOGLE, and a pale pink rectangle fills the background behind the right side. Printed on the image: 57% Hydrogen's Terminal-Bench 4.1 score vs 18.2% for previous Gemini 3.8 Flash; Google's own test.

Google DeepMind released Gemini 4 Hydrogen on November 12, positioned as the fast, cheap model in the Gemini 4 family and aimed at coding and agent work (Google DeepMind, 2026-11-12). People coding in Google's tools no longer have to pick one model themselves. The tools send deep reasoning to Gemini 4 Carbon, Google's larger and pricier model, and execution to Hydrogen, so cost depends on how much work goes to each.

Hydrogen has a single price across its full 1-million-token context: US$1.50 per million input tokens and US$7.50 per million output tokens. Hydrogen went live the same day in the Gemini API, AI Studio, the Gemini app and Google Search.

Most of the comparisons with the previous generation are Google's own tests. Terminal-Bench 4.1 measures engineering tasks completed in a terminal. In Google's test Hydrogen scored 57%, against 18.2% for the previous fast model, Gemini 3.8 Flash.

Orchestrator, a new feature in Antigravity, Google's development environment, divides the work between the two models. Within one work session, Orchestrator switches automatically between Gemini 4 Carbon for deep reasoning and Hydrogen for execution. Google writes that on Terminal-Bench 4.0, Orchestrator's success rate comes close to Carbon's while using 24% less usage quota on average and running 19% faster. Google gives no figure for how close, and the benchmark version differs from the 4.1 above.

Not proven yet: Google's same-class comparison names no other company's model How we checked →

Next to watch: third-party leaderboard scores comparing Hydrogen with other companies' models at the same price point.

Source and checks

This item ran in the brief of Nov 13, 2026, alongside 16 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.

Cite this item

SecondSource (2026). "Google launches the low-cost model Gemini 4 Hydrogen, which scores 57% on the Terminal-Bench 4.1 terminal-task benchmark in Google's own test, against 18.2% for the previous Gemini 3.8 Flash." https://secondsource.io/en/news/2026-11-13-2/. Accessed access date.