Gemini 4 Hydrogen Goes Live in Google Search Worldwide

Google launched Gemini 4 Hydrogen globally in Search and the Gemini app, claiming 2.5x faster tasks and stronger benchmark scores than Argon.

@sharedot

Gemini 4 Hydrogen Goes Live in Google Search Worldwide

By @sharedot · · 6 pages

  • AI
  • Google Gemini
  • Gemini 4 Hydrogen

Google launched Gemini 4 Hydrogen globally in Search and the Gemini app, claiming 2.5x faster tasks and stronger benchmark scores than Argon.

Google Ships Hydrogen to Every Search User

Google has released Gemini 4 Hydrogen, rolling the speed-focused model out globally from today in Google Search for all users, with selection also available in the Gemini app's model dropdown.

Tech Times calls it Google's biggest model improvement to Search yet, and Search's AI features previously ran on Gemini 3.8 Flash, which launched in September.

DeepMind senior vice president Koray Kavukcuoglu wrote that Hydrogen is 'built for rapid execution across diverse agentic workflows,' while the frontier Carbon and Argon models excel at deep, long-horizon reasoning.

The rollout also reaches all tiers of Google AI Studio, the Gemini API, and the Antigravity platform for paid AI Pro and Ultra subscribers, according to Tech Times.

A Fast Model That Beats the Frontier Flagship

The surprise is that a model built for speed is also winning on intelligence.

According to Tech Times, Hydrogen reportedly outperformed the more powerful Gemini 4 Argon on the Artificial Analysis Intelligence Index at its default high thinking level, and Google itself claims Hydrogen surpasses Argon when using that default setting, as Mashable reports.

Tech Times also reports Google claims Hydrogen completes tasks 2.5 times faster than its predecessor, Gemini 3.8 Flash. Google positions Hydrogen as an everyday execution engine that can pair with Carbon and Argon as the high-efficiency worker in multi-model workflows rather than a lesser sibling.

Search Feels More Like a Conversation

In Search, Hydrogen is designed to understand complicated multi-part questions, interpret user intent, and cope with confusing spelling mistakes, so people can type questions as they think of them instead of crafting keyword phrases.

Mashable reports Hydrogen responds with more visual answers and a wider variety of relevant links, and follows users' trains of thought including typos.

Tech Times adds that the model improves Google's search fan-out technique, which breaks one question into related subqueries to surface websites a narrow reading would miss, and that AI Mode can now return images, comparison tables, and custom interactive tools depending on the request.

Safety Trade-offs and Agentic Stakes

For agentic work, Google says Hydrogen can write code, run tests, and verify code changes, and Mashable notes it is intended for everyday work rather than designing systems or fixing complex bugs.

Tech Times reports Hydrogen ranks fifth on Agent Arena benchmarks involving multi-step activities such as deep research, data analysis, and navigating large codebases.

On safety, Google claims Hydrogen is two times less likely to circumvent permission boundaries than 3.8 Flash, per Mashable, but the company says it is still working on jailbreak resistance and cautions that Hydrogen remains prone to hallucinations like all generative AI agents.

The four thinking levels — low, medium, high and a new xhigh tier — trade off quality, latency and cost.

Sources

  1. mashable.com › Google unveils Gemini 4 Hydrogen, its new AI model built for speed
  2. techtimes.com › Google Introduces Gemini 4 Hydrogen to Make Search More Intelligent and Responsive

More on AI