Google Releases Gemini 4 Hydrogen Model for Handling Complex Questions
Technology company Google has introduced Gemini 4 Hydrogen, an AI model designed to answer multi-part and highly specific questions. It is already available in Google Search, the Gemini app and developer tools.
Technology company Google has released a new artificial intelligence model, Gemini 4 Hydrogen, which it says handles complex questions better, including those with several parts, hard-to-understand typos or highly specific requests. The model is available from the day of the announcement in Google Search, the Gemini app, all tiers of Google AI Studio, the Gemini API, the business version of Gemini, and Google Antigravity for Google AI Pro and Ultra subscribers.
The model can break a question into parts and address each one in its response. Robby Stein, vice president of product at Google Search, said it has become the new default model in AI Mode and is used for the hardest questions in AI Overviews.
More links and visuals
Google says Gemini 4 Hydrogen provides a wider range of links in its answers by running simultaneous queries across all relevant subtopics. Responses also include more visuals, such as images, comparison tables, graphs and maps. In one of Stein's examples, the AI built an interactive animated driving simulation in answer to a question about how phones work out which highway lane a car is in where there is no cell service.
The company also points to improved language understanding across all languages supported by AI Mode.
New thinking level
Users get more thinking-level options: low, medium and high, plus a new "xhigh" mode. According to Google, compared with Gemini Flash 3.8 the new model delivers significant leaps in both speed and intelligence, and in its class it matches or outperforms some competing models, such as Claude Sonnet 5.5 and GPT-6.1 Terra.
