Google releases Gemini 4 Hydrogen: 2.5 times faster than Gemini 3.8 Flash, and its overall benchmark score surpasses Argon.

Beating AI News Flash: Google has launched Gemini 4 Hydrogen, succeeding the previous Gemini 3.8 Flash, targeting daily programming and AI agent tasks. In Artificial Analysis’ latest comprehensive intelligence index, Hydrogen scores 51 points in its default high reasoning mode, surpassing the prior flagship Argon’s 50 points. Switching to maximum reasoning mode lifts the score to 54 points, though processing time increases significantly. Per Google’s internal testing, Hydrogen completes tasks 2.5 times faster than the 3.8 Flash. In the Terminal-Bench 4.1 programming benchmark, its score jumps from 18.2% (the previous generation’s result) to 57%, yet it still trails Claude Sonnet 5.5’s 63.9%. Google also added an Orchestrator function to Antigravity, which automatically switches models within a single task. For instance, it can assign Carbon to handle complex architecture design and challenging problems, then pass the work to Hydrogen for coding and test runs. Google’s tests found this combination delivers success rates nearly matching Carbon’s standalone performance, while cutting usage quota by 24% and boosting task completion speed by 19%. Hydrogen is now available to users of Gemini apps, Google Search, and AI Studio, while the Antigravity Orchestrator feature is exclusive to Pro and Ultra subscribers. API pricing stands at $1.5 per million input tokens and $7.5 per million output tokens—twice the current promotional rate of the 3.8 Flash.