Google has introduced Gemini 4 Hydrogen, the latest addition to its advanced AI model lineup, specifically designed for significantly faster task execution. This new model is now being rolled out globally, integrating directly into Google Search, the Gemini app, developer tools, and enterprise services.
Gemini 4 Hydrogen is engineered to handle a wide array of practical applications, from intricate coding and debugging to comprehensive research and data analysis. It excels at multi-step workflows and can process diverse content formats, including PDFs, spreadsheets, charts, audio, and recorded videos, making it a versatile tool for both individual users and businesses.
Enhanced Performance and Features
According to Google, Gemini 4 Hydrogen is 2.5 times faster than its predecessor, Gemini 3.8 Flash, when operating at its default high-thinking setting. Performance benchmarks further highlight its capabilities, with Gemini 4 Hydrogen scoring 57% on Terminal-Bench 4.1 for terminal-based tasks, a substantial leap from Gemini 3.8 Flash's 18.2%.
The model offers four distinct thinking levels—low, medium, high (default), and extra high (xhigh)—allowing users to tailor its processing intensity to specific needs. Furthermore, Gemini 4 Hydrogen is a key component of Google's multi-model workflow, orchestrated via its Antigravity platform. This system intelligently switches between models like Hydrogen and Carbon, optimizing task success rates while reducing resource consumption by an average of 24% and increasing completion speed by 19%.
Availability and Pricing
Gemini 4 Hydrogen is accessible to a broad audience. Free users of the Gemini app will benefit from its capabilities through automatic model selection, while paid subscribers can manually choose it via the model selector. Developers can integrate the model using the Gemini API and Google AI Studio, and businesses can leverage it through Gemini for business. Eligible Google AI Pro and Ultra subscribers also have access via Google Antigravity.
For developers utilizing the Gemini API, pricing is set at $1.50 per million input tokens and $7.50 per million output tokens, consistent across its one-million-token context window. Google emphasizes that Gemini 4 Hydrogen also incorporates improved safeguards to minimize incorrect or unintended actions during tool usage and task completion.