Google has released Gemini 4 Hydrogen, which it says can break down complicated questions with multiple parts, hard-to-understand typos or highly specific requests.
The model became available on Thursday in Search, the Gemini app, all tiers of Google AI Studio, the Gemini API, Gemini for business and in Google Antigravity for Google AI Pro and Ultra subscribers.
Robby Stein, vice president of product at Google Search, said the model is now the new default in AI Mode and is being used for the hardest questions in AI Overviews.
According to Google, Gemini 4 Hydrogen delivers significant leaps in both speed and intelligence compared with Gemini Flash 3.8, and matches or outperforms some competing models in its class, such as Claude Sonnet 5.5 and GPT-6.1 Terra.
Users have a wider choice of thinking levels: low, medium, high and a new “xhigh” mode.
The company said the model generates responses with more visuals, including images, comparison tables, graphs and maps. It also said the model conducts simultaneous queries across all relevant subtopics to provide a wider range of links.
One of Stein’s examples showed the AI generating an interactive animated driving simulation in response to a question about how phones locate which lane of a highway a car is on in areas without cell service.
Google said the new model also has improved language understanding across all supported AI Mode languages.
Analysis
Why This Matters
- The model is immediately available as the default in Google’s AI-powered search features, making its reasoning capabilities part of everyday search for billions of users
- Multi-part question handling and visual responses could shift how users interact with search engines, reducing the need to refine queries
- The comparison with Claude Sonnet 5.5 and GPT-6.1 Terra signals an intensifying AI model competition, with implications for product pricing and capabilities
Background
Google has been rolling out generative AI features in Search since 2023, including AI Overviews and AI Mode. The company regularly updates its Gemini family of models. Gemini 4 Hydrogen is the latest iteration, following Gemini Flash 3.8.
Key Perspectives
Google (Robby Stein, VP of product at Google Search): Calls Gemini 4 Hydrogen “the biggest model upgrade to Search yet” and says it is “now the new default model in AI Mode & being used for your hardest questions in AI Overviews.”
Google (company statement): Says the new model “delivers significant leaps in both speed and intelligence” compared with the previous version and “matches or outperforms some of the competing models in its class,” specifically naming Claude Sonnet 5.5 and GPT-6.1 Terra.
What to Watch
- How users respond to the new “xhigh” thinking mode in practice
- Whether performance claims about matching competitors are borne out in independent benchmarks
- Adoption rate among developers using the Gemini API and Google AI Studio
Checked
11 claims in this story checked against 1 source: 10 supported.
Show each claim
- SupportedStein says Gemini 4 Hydrogen is Google Search’s biggest model upgrade yet.The Verge: “Google’s Robby Stein says Gemini 4 Hydrogen is ‘the biggest model upgrade to Search yet.’”
- SupportedGemini 4 Hydrogen can address questions with multiple parts, hard-to-understand typos or highly specific requests.The Verge: “Google says its new Gemini 4 Hydrogen model can give more helpful answers to complex questions, including “questions with multiple parts, hard-to-understand typos, or highly specific requests.””
- SupportedThe model’s responses include more visuals.The Verge: “Its responses have more visuals, too, like images, comparison tables, graphs, and maps.”
- Not in the sourcesGemini 4 Hydrogen became available on Thursday.The Verge: “It’s available starting today in search, the Gemini app, all tiers of Google AI Studio, the Gemini API, Gemini for business, and in Google Antigravity for Google AI Pro and Ultra subscribers.”
- SupportedThe model is available in Search, the Gemini app, all tiers of Google AI Studio, the Gemini API, Gemini for business, and Google Antigravity for Google AI Pro and Ultra subscribers.The Verge: “It’s available starting today in search, the Gemini app, all tiers of Google AI Studio, the Gemini API, Gemini for business, and in Google Antigravity for Google AI Pro and Ultra subscribers.”
- SupportedStein said the model is the new default in AI Mode and is being used for the hardest questions in AI Overviews.The Verge: “Robby Stein, VP of product at Google Search, says it’s “now the new default model in AI Mode & being used for your hardest questions in AI Overviews.””
- SupportedGoogle says Gemini 4 Hydrogen makes significant speed and intelligence gains over Gemini Flash 3.8 and matches or outperforms some models in its class, including Claude Sonnet 5.5 and GPT-6.1 Terra.The Verge: “According to Google, the new model “delivers significant leaps in both speed and intelligence” compared to Gemini Flash 3.8, and it “matches or outperforms some of the competing models in its class,” like Claude Sonnet 5.5 and GPT-6.1 Terra.”
- SupportedUsers have low, medium and high thinking levels, plus a new “xhigh” mode.The Verge: “Users have more thinking level options with Gemini 4 Hydrogen, as well, including low, medium, high, and a new “xhigh” mode.”
- SupportedGoogle says the model conducts simultaneous queries across relevant subtopics to provide a wider range of links.The Verge: “Google says Gemini 4 Hydrogen also gives users a wider range of links in its responses by “conducting simultaneous queries across all the relevant subtopics.””
- SupportedOne of Stein’s examples shows an interactive animated driving simulation about how phones locate a highway lane without cell service.The Verge: “One of Stein’s examples shows the AI generating an interactive animated driving simulation in response to a user’s question about how phones locate which lane of a highway a car is on in areas without cell service.”
- SupportedGoogle says the model has improved language understanding across all supported AI Mode languages.The Verge: “Google says the new model also has improved language understanding “across all supported AI Mode languages.””