Google's new everyday AI just took over Search

Reviewed byNidhi Govil

11 Sources

Share

Google's new everyday AI model, Gemini 4 Hydrogen, launched November 12 and now powers Search worldwide. It replaces Flash, and every benchmark behind it comes from Google itself.

Google launched Gemini 4 Hydrogen on November 12, a fast, efficiency-focused model for everyday coding and agentic work that now runs Google Search for every user worldwide

5

4

. It also replaces Flash, the everyday model Google shipped two months ago, in the Gemini app

2

1

.

Source: Google

Source: Google

Google, the maker of the Gemini model family, pitches Hydrogen as bringing "cutting-edge intelligence to everyone -- with speed and efficiency for everyday coding and agentic workflows"

5

. It sits alongside the frontier pair Gemini 4 Carbon and Gemini 4 Argon, launched weeks earlier, and arrives days after Carbon itself became generally available

5

2

3

.

An everyday model that outscored a frontier one

In Google's own testing on a single third-party index, the Artificial Analysis Intelligence Index, Hydrogen scored 51 against Argon's 50 at its default high thinking setting

5

2

. It reaches 54 at the new 'xhigh' level

5

2

. Digital Trends, the one outlet to read Google's benchmark tables, finds Hydrogen ahead of OpenAI's GPT 6.1 Terra and of Anthropic's Claude Sonnet 5.5 on knowledge and multimodal tests

3

. It trails them slightly on science and math, computer use and Terminal-Bench 4.1

3

.

Against its predecessor Gemini 3.8 Flash, launched September 2, Google claims Hydrogen finishes tasks 2.5 times faster

5

3

. It also says the new model more than triples Flash's Terminal-Bench 4.1 score and cuts fabricated answers more than fivefold

5

3

. Google also publishes a long list of safety claims. It reports agentic overreach down by up to 81%, misaligned actions and shortcutting cut twelvefold, and a 96.7% score on its internal agent honesty evaluation

5

3

.

Pricing and where you find it

Pricing is flat across the model's 1 million-token context window: $1.50 per million input tokens, $7.50 per million output tokens

5

6

2

. Hydrogen is available across every tier of Google AI Studio and the Gemini API, in the Gemini app, and in Google Search

5

6

3

. It also runs in the Antigravity coding tool for paid AI Pro and Ultra subscribers

5

3

. Free Gemini app users get it automatically when the app decides it fits the task

5

.

In the app's model picker it is labelled "Everyday help", sitting between the quick-answer Flash-Lite and the research-oriented 3.1 Pro

2

. A writer at XDA Developers found Flash gone from their own model picker, though Google has not formally said the Flash line is discontinued

1

. Digital Trends reports it is also coming to Gemini in Google Sheets for paid subscribers

3

.

Source: 9to5Google

Source: 9to5Google

Alongside the launch, Google introduced Orchestrator in Antigravity, which switches between models such as Carbon and Hydrogen within a single session. Per Google, it came close to Carbon's task-success rate on one benchmark, Terminal-Bench 4.0, while using 24% less quota and finishing 19% faster

5

2

. In partner tests on data-heavy enterprise work, Google says Hydrogen ran up to 2.8 times faster while consuming up to 23% fewer tokens

3

.

Search, at full scale

Google calls Hydrogen "the biggest model upgrade to Search yet"

4

, and the rollout is global

4

3

. The claimed gains cover multi-part questions and typos, more visual answers such as images or tables, and stronger language understanding

4

3

.

Source: Google

Source: Google

Every performance, safety and efficiency figure in this story is a Google claim, from its own benchmarks and partner tests; no independent testing appears in any of the coverage

5

3

.

© 2026 TheOutpost.AI All rights reserved