Google has introduced Gemini 4 Hydrogen, adding another model to its Frontier series alongside the Gemini 4 Carbon and Gemini 4 Argon within a few days.
Designed to quickly execute agent workflows, Hydrogen brings Google’s latest modeling capabilities to everyday coding, software development, and other tasks that require speed and efficiency.
Unlike Carbon and Argon, which are designed for deep, long-term thinking about complex problems, Hydrogen focuses on rapid execution. It can write code, run tests, review changes, and perform routine development tasks, either independently or alongside the other Gemini 4 models.
Hydrogen also outperformed Argon on the Artificial Analysis Intelligence Index, with a score of 51 versus Argon’s 50 at the standard high level of reasoning. According to Google, the score rises to 54 at the highest xhigh setting.
Faster coding and smarter execution
Google designed Hydrogen to handle everyday developer workloads, including debugging, repository coding, web development, and multi-stage agent tasks. It can also handle text, images, audio and video, with a combined input-output context window of up to 1 million tokens.
The model offers four levels of thinking: low, medium, high, and a new xhigh setting that allows developers to adjust the depth of thinking according to the complexity, latency, and cost requirements of a task.
In Google reviews, Hydrogen scored 57 percent on Terminal Bench 4.1, more than three times Gemini 3.8 Flash’s 18.2 percent. The company also claims that Hydrogen completes tasks 2.5 times faster than its predecessor at the default high setting.
For more comprehensive knowledge work, the model can analyze documents, spreadsheets, charts and recorded videos to produce reports and structured output. Google also highlights improvements in tool usage, coding accuracy, and its ability to recover when a tool call fails or a search returns no results.
Orchestration of multiple models
Google is also introducing Orchestrator in its Antigravity development platform, allowing developers to combine Gemini 4 Carbon and Hydrogen in a single session.
Instead of relying on one model during a task, Orchestrator dynamically switches between the two depending on the work required. Carbon can initially examine a codebase, resolve ambiguities, and create an execution plan, while Hydrogen handles routine coding, package updates, and testing. If the faster model encounters difficulties, the task can be returned to Carbon for further consideration and adjustment.
According to Google, Orchestrator on Terminal-Bench 4.1 achieved task success rates close to those of Carbon, using 24 percent less usage quota and completing tasks an average of 19 percent faster.
The approach is intended to reduce the need to use a high-reasoning model for each step of a development workflow, potentially reducing costs while maintaining performance.
Google is introducing hydrogen to Google AI Studio, the Gemini API, Gemini for Business, the Gemini app and Google Search. It is also available to paid Google AI Pro and Ultra subscribers in Antigravity, while Orchestrator is initially rolling out to Google AI Ultra subscribers.
The model’s API price is $1.50 per million input tokens and $7.50 per million output tokens in its 1 million token context window.
