Google says new Gemini model improves understanding of complex questions

Google has announced the release of its new Gemini 4 Hydrogen model, which is designed to provide improved assistance for complicated queries. The company states that the technology can now handle multipart questions, requests containing difficult typos, and highly specific inquiries by breaking them down into individual components.

According to the firm, this model is available from today across several platforms, including Google Search, the Gemini app, Google AI Studio, the Gemini API, Gemini for business, and Google Antigravity for Google AI Pro and Ultra subscribers. Robby Stein, VP of product at Google Search, confirmed that it is “now the new default model in AI Mode & being used for your hardest questions in AI Overviews.”

Gemini 4 Hydrogen is capable of providing a broader selection of links by “conducting simultaneous queries across all the relevant subtopics.” Furthermore, the outputs now feature more visual elements, such as graphs, images, comparison tables, and maps. Robby Stein demonstrated the capability by showing the AI generate an interactive animated driving simulation to explain how phones detect highway lanes in areas lacking cell service. Google reports that language comprehension has been enhanced “across all supported AI Mode languages.”

Users can now access various thinking level settings, including low, medium, high, and a new “xhigh” mode. Google claims the model “delivers significant leaps in both speed and intelligence” when compared to the Gemini Flash 3.8 version. The company further states that it “matches or outperforms some of the competing models in its class,” specifically citing Claude Sonnet 5.5 and GPT-6.1 Terra.