Google Introduces Gemini 4 Hydrogen Model for Coding and Agentic Workflows

What Happened

On 12 November 2026, Google announced the introduction of Gemini 4 Hydrogen, a model designed for speed and efficiency in everyday coding and agentic workflows [1, 2]. While Google’s frontier models, Gemini 4 Carbon and Gemini 4 Argon, focus on deep, long-horizon reasoning, Hydrogen is built for rapid execution [1].

Key technical details and performance metrics include:

• Hydrogen features four thinking levels: low, medium, high (the default), and a new xhigh tier [1, 3].
• On the Artificial Analysis Intelligence Index, Hydrogen scored 51 at its default high thinking level and reached 54 at its xhigh setting, surpassing Argon’s score of 50 [1].
• Compared to Gemini 3.8 Flash, Hydrogen finishes tasks 2.5 times faster at its high default setting and more than triples Terminal-Bench 4.1 scores, moving from 18.2% to 57% [1].
• It also reduces fabricated claims by more than 5x [1].
• The model is priced at $1.50 per million input tokens and $7.50 per million output tokens across its 1 million-token context window [1].
• Google is also introducing Orchestrator in Google Antigravity, which dynamically switches between models like Carbon and Hydrogen. On Terminal-Bench 4.0, Orchestrator reached near-Carbon task success rates while using 24% less quota and completing tasks 19% faster on average [1].

Google Search is also being upgraded. Robby Stein, VP of product at Google Search, stated that Hydrogen is the new default model in AI Mode and is used for difficult questions in AI Overviews [3]. He described it as the biggest model upgrade to Search yet [3]. The model can handle complex questions with multiple parts or typos and provides more visuals, such as comparison tables, graphs, and maps [3].

Who Is Affected

Gemini 4 Hydrogen is available for several groups of users:

• Developers can access it via Google AI Studio, the Gemini API, and Android Studio [1, 2].
• Businesses can use it through Gemini for business [1, 2].
• General consumers can use it in the Gemini app and through Google Search [1, 3].
• Paid Google AI Pro and Ultra subscribers can access it in Google Antigravity [1, 2].

According to Google, the model matches or outperforms competing models in its class, such as Claude Sonnet 5.5 and GPT-6.1 Terra [3].