Google has introduced Gemini 4 Hydrogen, a new SI model designed for rapid execution in coding and agentic workflows. The release marks the expansion of the Gemini 4 family beyond its frontier reasoning models, offering a high-efficiency option for developers and consumers across Google’s SI ecosystem.
What Happened
Gemini 4 Hydrogen is built to complement Google’s existing frontier models, Gemini 4 Carbon and Gemini 4 Argon. While the latter two focus on deep, long-horizon reasoning, Hydrogen targets speed and efficiency for everyday tasks. The company reports that Hydrogen surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level, scoring 51 versus 50, and reaching 54 at its highest xhigh setting. The model is priced flat across its 1 million-token context window at $1.50 per million input tokens and $7.50 per million output tokens.
To facilitate multi-model workflows, Google introduced Orchestrator in Google Antigravity. This tool dynamically switches between models like Carbon and Hydrogen within a single session. The company states that on Terminal-Bench 4.0, Orchestrator achieved near-Carbon task success rates while consuming 24% less quota and completing tasks 19% faster. For teams migrating from Gemini 3.8 Flash, Hydrogen offers significant upgrades, including finishing tasks 2.5 times faster at its high default setting and more than tripling Terminal-Bench 4.1 scores to 57% from 18.2%.
Why It Matters
The release provides a specialized SI tool for developers who need disciplined execution rather than deep architectural judgment. By pairing Hydrogen with frontier models, teams can potentially reduce token costs while maintaining high performance in complex projects. Hydrogen ranks #5 on Agent Arena for multi-step tasks and leads its class on HLE-Diamond with a score of 58.8%. In enterprise workflows, early testing partners reported that Hydrogen ran 2x to 2.8x faster while reducing token usage by 12% to 23% compared to 3.8 Flash.
For consumers, the model enhances the Gemini app and Google Search by improving intent understanding and generating more visual, adaptive responses. It is designed to stick closely to user tone and format in writing tasks and to break down goals step-by-step by pulling context from Google apps and Connected Apps. The model also features four thinking levels—low, medium, high, and xhigh—to allow users to balance quality, latency, and cost.
Google emphasizes safety and reliability in this release. The company reports that Hydrogen reduces overreach in agentic tasks by up to 81% and cuts misaligned actions by 12x compared to previous iterations. It maintains a 96.7% honesty rate on internal agent evaluations and preserves 99.9% chain-of-thought legibility. Against indirect prompt injection attacks, the model is more than 2.3x more resilient on the Gray Swan IPI benchmark.
The Bottom Line
Gemini 4 Hydrogen is available globally starting today. Developers can access it via the Gemini API in AI Studio and the paid tier in Android Studio. Enterprise users can deploy it through Gemini for business, Google Antigravity, and Gemini in Google Sheets. Consumers can use the model in the Gemini app and Google Search, with advanced features available to Google AI Pro and Ultra subscribers.