Google Launches Gemini 4 Hydrogen as Its Fastest AI Model Yet in Search AI Mode

Google has launched Gemini 4 Hydrogen, its fastest artificial intelligence model yet, and made it the default engine behind Search's AI Mode for users around the world. Rolled out on Thursday, November 12, 2026, the new model completes tasks 2.5 times faster than its predecessor, Gemini 3.8 Flash. It also powers the hardest questions in AI Overviews, the AI summaries at the top of Google Search results. Google says Hydrogen was built for everyday tasks, coding and agentic workflows. It is available across Google AI Studio, the Gemini API, Gemini for business, the Gemini app and Google Search, and in Google Antigravity for Pro and Ultra subscribers. Pricing is $1.50 per million input tokens and $7.50 per million output tokens, across a one-million-token context window.

Key Points

• Google rolled out Gemini 4 Hydrogen on Thursday, November 12, 2026, and made it the default model in Search AI Mode worldwide.
• The model completes tasks 2.5 times faster than its predecessor, Gemini 3.8 Flash.
• It offers four thinking levels — low, medium, high and extra-high — so users pay only for the reasoning a task needs.
• It also handles the hardest questions in AI Overviews, and is available across AI Studio, the Gemini API, Gemini for business, the Gemini app and Google Search.
• Pricing is $1.50 per million input tokens and $7.50 per million output tokens, with a one-million-token context window.

What Happened

Google rolled out Gemini 4 Hydrogen on Thursday, November 12, 2026, Pacific time, and switched it on as the default model in AI Mode, the conversational search experience inside Google Search, for all users worldwide. Robby Stein, Google's vice president of product for Search, said the model is now the standard choice in AI Mode and is also used for the hardest questions in AI Overviews. People asking Google's AI tools their hardest questions will now get answers from Hydrogen by default.

The launch is a speed play. Google says Hydrogen completes tasks 2.5 times faster than Gemini 3.8 Flash. That matters most where waiting is the weakness: long research questions in AI Mode, multi-step coding jobs, and agentic workflows. A model that finishes each step in less than half the time can attempt more steps and still answer sooner. Availability is broad from day one, with Hydrogen live in AI Studio, the Gemini API, Gemini for business, the Gemini app and Google Search.

Background: The Gemini 4 Family

Hydrogen is the third member of Google's Gemini 4 family to reach the public, and each member has a different job. Argon, announced earlier, is aimed at complex engineering and research, where depth matters more than speed. Carbon is built for long-running tasks that unfold over hours rather than minutes. Hydrogen completes the set as the fast, efficient model for everyday use, the one most people will meet, because it now sits behind Search.

The launch comes weeks after Google said the Gemini app had passed one billion users. Putting its newest model at the heart of AI Mode and AI Overviews is the fastest way to put it in front of that audience. It also raises the stakes: any weakness in Hydrogen will now show up in the answers Search gives to its hardest questions.

Gemini 4 Hydrogen: Speed, Thinking Levels and Pricing

The headline number is the 2.5 times speed improvement over Gemini 3.8 Flash. Google has framed Hydrogen as a leap in how fast reasoning is delivered, built for speed and efficiency. For users, that should mean shorter waits in AI Mode and snappier answers in the Gemini app, and developers should see lower latency in products built on the Gemini API.

The model's four thinking levels — low, medium, high and extra-high — are its most distinctive feature. Instead of one fixed amount of reasoning for every question, Hydrogen lets the task decide how hard the model thinks. A simple question can run at the low level and come back almost instantly, while a hard research or coding problem can run at extra-high. Users only wait, and developers only pay, for the depth the task actually needs. Pricing is $1.50 per million input tokens and $7.50 per million output tokens, with a one-million-token context window.

Reactions and Quotes

Google chief executive Sundar Pichai announced the launch in a post on X, saying the model was built for speed and efficiency. Stein said Hydrogen is now the default model in AI Mode for all users worldwide and is used for the hardest questions in AI Overviews. The Search leaders fronting the announcement show where Google expects Hydrogen to matter most: in everyday questions, not laboratory benchmarks. Industry watchers will now look for independent benchmarks on real tasks, and for developer feedback on whether the four thinking levels deliver the savings Google promises.

Why It Matters

Making Hydrogen the default in AI Mode matters because defaults decide what most people actually use. Few Search users change model settings, so the model Google chooses becomes, for practical purposes, the AI that answers the world's hardest search questions. If Hydrogen is as fast as Google claims, those answers will arrive sooner and AI Mode will feel noticeably sharper, drawing more people toward conversational answers.

It also matters for the economics of AI. A model that is 2.5 times faster can serve more than twice as many users on the same hardware, and the thinking levels let Google match cost to difficulty instead of charging every question at the top rate. For developers, that could make Hydrogen the obvious choice for high-volume products. For Google, it protects the economics of giving AI answers away free inside Search. For more on the pressures facing the AI giants, see our report on nonprofits suing OpenAI, Google, Anthropic, Meta and SpaceX to freeze frontier AI.

What Happens Next

The first test will be usage. Google will be watching how AI Mode and AI Overviews perform on Hydrogen, and developers will test the four thinking levels against real workloads, with verdicts showing up in benchmarks over the coming weeks. More detail is likely to follow, including performance figures and any expansion of Hydrogen into more Google products. For now, the model is live and the default in Search AI Mode. Readers can also follow the wider AI hardware race in our coverage of IBM's Nighthawk quantum chip and Jensen Huang's resignation as Nvidia chief executive.

Frequently Asked Questions

What is Gemini 4 Hydrogen?
Google's newest AI model, launched November 12, 2026: the fast, efficient member of the Gemini 4 family, now the default in Google Search's AI Mode.

How much faster is it than Gemini 3.8 Flash?
Google says Hydrogen completes tasks 2.5 times faster than its predecessor, Gemini 3.8 Flash.

Where is Gemini 4 Hydrogen available?
In Google Search's AI Mode and AI Overviews, Google AI Studio, the Gemini API, Gemini for business and the Gemini app.

What are the four thinking levels?
Low, medium, high and extra-high. They let the model spend more effort on hard tasks and less on simple ones.

How much does Gemini 4 Hydrogen cost?
$1.50 per million input tokens and $7.50 per million output tokens, with a one-million-token context window.