Google Launches Gemini 4 Hydrogen, Its Fastest AI Model for Search and Coding
Google has launched Gemini 4 Hydrogen, a new artificial intelligence model built for speed, and it is rolling out across Google Search, the Gemini app and Google’s developer tools starting Thursday, November 12. Hydrogen replaces Gemini 3.8 Flash as Google’s fast everyday model, and Google says it completes tasks 2.5 times faster than its predecessor while making fewer mistakes.
The launch matters most for ordinary users because Hydrogen is now the model behind Google Search for everyone. Google says it is the biggest upgrade to Search yet, with better answers to complicated questions, more visual results and links to a wider range of websites.
“Today, we’re introducing Gemini 4 Hydrogen, bringing cutting-edge intelligence to everyone, with speed and efficiency for everyday coding and agentic workflows,” Google said in its announcement.
Table of Contents
What Is Gemini 4 Hydrogen
Gemini 4 Hydrogen is the fast, low-cost member of Google’s Gemini 4 family. While the frontier models in the family, Gemini 4 Carbon and Gemini 4 Argon, handle deep, long-horizon reasoning on the hardest problems, Hydrogen is built for rapid execution across everyday and multi-step tasks: writing code, running tests, doing research, analyzing data and building web pages.
Google is bringing Gemini 4 to everyone through Hydrogen. It is available now for all tiers of Google AI Studio, the Gemini API, Gemini for business, the Gemini app, Google Search and paid Google AI Pro and Ultra subscribers in Google Antigravity, Google’s agentic coding tool.
The model comes with four thinking levels: low, medium, high and a new xhigh tier. High is the default. Google says the levels let users trade speed against reasoning depth, so they only wait and pay for as much thinking as a task needs.
How Fast Is It, and How Good
Google’s headline numbers, all from the company’s own testing, show a large jump over Gemini 3.8 Flash. Hydrogen finishes tasks about 2.5 times faster at its default high thinking setting, and its score on Terminal-Bench 4.1, a coding benchmark, rose to 57 percent from 18.2 percent for 3.8 Flash. Google also says Hydrogen cuts fabricated claims by more than five times and adapts cleanly when a tool call fails or a search returns no results.
On the Artificial Analysis Intelligence Index, Hydrogen scores 51 at its default setting, one point above Gemini 4 Argon, and reaches 54 at its highest xhigh thinking level. It also ranks fifth on Agent Arena, a live benchmark for multi-step agent tasks, and leads its class on several specialist tests, including HLE-Diamond for deep reasoning at 58.8 percent.
Independent testing will show how well Google’s figures hold up in real use. Benchmarks measure performance on fixed tests, not on every prompt a user will try, so the practical difference for most people will be speed and reliability on routine work rather than any single score.
What Changes in Google Search
For most users, the biggest change is in Search. Hydrogen is now the default model in AI Mode, Google’s conversational search experience, and it handles the harder questions in AI Overviews. Google says Search responses will become more visual and easier to scan, with a wider range of relevant links.
The model is designed to understand questions written the way people naturally think them, even when they have several parts, contain typos or carry very specific instructions. Google is also upgrading its query fan-out technique, which breaks a question into parallel searches across its component topics, so Search can connect users to a broader set of helpful links, including sites they might not have discovered before.
That last point matters for publishers and website owners. More links in AI answers could mean more discovery for smaller sites, but the real test will be referral traffic and whether the work is represented accurately, not how many citations appear on a demo screen. Google has not supplied publisher traffic figures for the rollout.
Free users get Hydrogen in Search for all users globally, and in the Gemini app through automatic model selection, where Gemini picks Hydrogen when it suits the prompt.
For Developers and Businesses
Hydrogen is also a major upgrade for teams moving from Gemini 3.8 Flash. Google says partners testing it on complex coding, SQL and data-heavy enterprise work saw it run two to 2.8 times faster while using 12 to 23 percent fewer tokens than 3.8 Flash.
The model pairs a 1 million-token input window with up to 1 million output tokens, so it can work across dense PDFs, spreadsheets, charts, audio and recorded video and turn them into cited reports and structured deliverables. It is priced flat across its full context window at $1.50 per million input tokens and $7.50 per million output tokens.
Google is also introducing Orchestrator in Google Antigravity, which switches between models like Carbon and Hydrogen inside one session. On Terminal-Bench 4.0, Orchestrator reached near-Carbon task success rates while using 24 percent less quota and finishing 19 percent faster. In multi-agent setups, Carbon can act as coordinator while Hydrogen powers worker and verifier agents that build and test changes at the same time.
Google’s AI Race

Google, led by chief executive Sundar Pichai, has been racing rivals to put its newest models into the products people already use, and Hydrogen is the clearest example of that strategy. Instead of keeping its best model behind a paywall, Google is putting a Gemini 4 model in front of every Search user at no extra cost, betting that speed and everyday usefulness will keep people inside Google’s ecosystem.
The launch comes a day after Google opened preorders for its first Gemini-powered smart glasses, which also run on the company’s AI assistant. Together, the two launches show Google pushing Gemini into Search, phones, developer tools and now eyewear in the same week.
Safety is part of the pitch too. Google says Hydrogen reduces overreach in agentic tasks by up to 81 percent and under-service by up to 79 percent, cuts misaligned actions and task shortcutting by 12 times in its internal evaluations, and is more than 2.3 times more resilient to indirect prompt injection attacks. It also reports a 96.7 percent honesty rate on an internal agent honesty evaluation.
What Happens Next

Hydrogen is rolling out globally now, but not every account will see it on day one. Google says the rollout covers Search, the Gemini app, AI Studio, the Gemini API, Gemini for business and Antigravity, with availability varying by product and account.
The real verdict will come from users. If Hydrogen is as fast and reliable as Google claims, most people will notice it first as quicker, clearer Search answers and fewer failed coding tasks. If the benchmarks flatter it, the gap between the announcement and everyday experience will show up quickly, because Search is the most used product Google has, and every query routed through Hydrogen is a live test.
FAQs
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is Google’s new fast, low-cost AI model, launched November 12, 2026. It replaces Gemini 3.8 Flash across Search, the Gemini app, AI Studio, the Gemini API and business tools.
Is Gemini 4 Hydrogen free to use?
Yes, for consumers. It is rolling out in Google Search for all users and in the free Gemini app through automatic model selection. Developers pay API rates of $1.50 per million input tokens and $7.50 per million output tokens.
How much faster is Gemini 4 Hydrogen than Gemini 3.8 Flash?
Google says Hydrogen completes tasks about 2.5 times faster than 3.8 Flash at its default high thinking setting, and its Terminal-Bench 4.1 score rose to 57 percent from 18.2 percent.
What is the difference between Gemini 4 Hydrogen, Carbon and Argon?
Carbon and Argon are Google’s frontier models for deep, long-horizon reasoning. Hydrogen is the fast execution model for everyday coding, research and multi-step tasks, and it is the one rolling out to everyone.
Where can I try Gemini 4 Hydrogen?
In Google Search for all users, in the Gemini app model menu, in Google AI Studio and through the Gemini API. Paid Google AI Pro and Ultra subscribers can also use it in Google Antigravity.