Skip to main content

Kaer NewsSoftware

Google launches Gemini 4 Hydrogen with faster answers and new thinking tier

Google released Gemini 4 Hydrogen on 12 November, claiming it completes tasks 2.5 times faster than Gemini 3.8 Flash and now powers Search AI Mode.

4 min read
Google launches Gemini 4 Hydrogen with faster answers and new thinking tierPhoto: The Verge

Key points

Google launched Gemini 4 Hydrogen, a faster and more intelligent model now default in Search AI Mode, with a new xhigh thinking tier.

Google launched Gemini 4 Hydrogen on 12 November, making it the new default model in Search AI Mode and available across the Gemini app, AI Studio, the API, and Antigravity. The company says the model completes tasks 2.5 times faster than Gemini 3.8 Flash at its default high setting, while delivering 'significant leaps in both speed and intelligence.' Robby Stein, VP of product at Google Search, called it 'the biggest model upgrade to Search yet.'

Hydrogen is designed for fast execution across agentic workflows, where an AI carries out multiple steps on its own, rather than the deep reasoning of frontier models like Gemini 4 Carbon and Argon. Google says using the most powerful model isn't always needed, and Hydrogen offers a faster, cheaper, and reasonably smart alternative. It introduces a new 'xhigh' thinking level alongside low, medium, and high, aimed at completing tasks faster with fewer output tokens.

Speed and benchmark gains

On benchmarks, Hydrogen surpasses Gemini 4 Argon on the Artificial Analysis Intelligence Index at its default high thinking level, scoring 51 versus 50, and reaches 54 at the highest xhigh setting. It leads its class on HLE-Diamond with 58.8%, ranks first on Vals Tax Agent Bench with 52.3%, and ties for first on the Vals Legal Research Bench with 55.8%. On multi-step tasks like deep research and data analysis, it ranks fifth on Agent Arena.

Compared to Gemini 3.8 Flash, Hydrogen more than triples Terminal-Bench 4.1 scores, rising from 18.2% to 57%, and cuts fabricated claims by more than 5x. It reduces overreach in agentic tasks by up to 81% and under-service by up to 79%, and cuts misaligned actions and task shortcutting by 12x. When faced with blocked tasks, it is up to 2x less likely to bypass permission boundaries, achieving a 96.7% honesty rate on Google's internal evaluation.

In Search, Hydrogen improves understanding of complex queries, including those with multiple parts, hard-to-understand typos, or highly specific requests. Its advanced reasoning enhances query fan-out, breaking a single prompt into multiple subqueries to surface a wider range of relevant links. Responses become more visual, with images, comparison tables, graphs, and custom tools, and language understanding improves across all supported AI Mode languages.

Search and app integration

The model is now the default in AI Mode and is used for 'your hardest questions' in AI Overviews, according to Stein. In the Gemini app, Hydrogen better follows tone, format, and rules, producing email drafts, study guides, and documents 'ready to use on the first try.' It also more frequently calls the right tools and skills, such as Connected Apps, for the job.

In Google Antigravity, Hydrogen offers an 'Extra' thinking tier alongside Low, Medium, and High, and is available to paid AI Pro and Ultra subscribers. A new Orchestrator feature dynamically switches between models like Carbon and Hydrogen within the same session, allowing Carbon to coordinate while Hydrogen powers worker and verifier agents. This multi-model collaboration aims to help teams ship complex projects faster without driving up token costs.

For developers, Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens. Google says partners have seen Hydrogen run 2x to 2.8x faster while reducing token usage by 12% to 23% compared to 3.8 Flash. The model is rolling out to all Search users globally, with availability in the Gemini app, all tiers of AI Studio, the Gemini API, and Gemini for business.

Availability and pricing

Hydrogen's release follows the availability of Gemini 4 Carbon earlier this week, and Google chose not to keep the 'Flash' branding for this model. The company says Hydrogen 'matches or outperforms some of the competing models in its class.' Google frames Hydrogen as bringing 'cutting-edge intelligence to everyone,' with the model picker in the Gemini app labeling it as providing 'Everyday help.'

Google says the update is rolling out to all Search users globally, with no specific completion date provided. Users can try the new capabilities by visiting Google Search and asking nuanced questions. The company says Hydrogen is available starting Thursday in search, the Gemini app, all tiers of Google AI Studio, the Gemini API, Gemini for business, and in Antigravity for AI Pro and Ultra subscribers.

Frequently asked questions

What is Gemini 4 Hydrogen?

Gemini 4 Hydrogen is a new AI model from Google launched on 12 November, designed for fast execution across agentic workflows. It is now the default model in Search AI Mode and available in the Gemini app, AI Studio, the API, and Antigravity.

How much faster is Gemini 4 Hydrogen than Gemini 3.8 Flash?

Google says Hydrogen completes tasks 2.5 times faster than Gemini 3.8 Flash at its default high setting. Partners have seen it run 2x to 2.8x faster while reducing token usage by 12% to 23%.

What is the new 'xhigh' thinking level?

The 'xhigh' thinking level is a new option in Gemini 4 Hydrogen, alongside low, medium, and high. It is designed to complete tasks faster with fewer output tokens, and at this setting Hydrogen reaches a score of 54 on the Artificial Analysis Intelligence Index.

More from this edition