Gemini 4 Hydrogen is Google fast model for agentic workflows, announced on 12 November 2026 as the successor to Gemini 3.8 Flash. Google describes it as a daily driver built for fast execution rather than deep long-horizon reasoning.
API pricing is 1.50 dollars per million input tokens and 7.50 dollars per million output, with cache hits at 0.075 dollars. Pricing is flat across the entire 1 million token context window rather than stepping up at a threshold, and up to 1 million output tokens are supported. Four thinking levels are available: low, medium, high as the default, and a new xhigh tier.
It scores 51 on the Artificial Analysis Intelligence Index at high effort and 54 at xhigh, against 50 for the larger Gemini 4 Argon. Against its own predecessor it triples Terminal-Bench 4.1 at 57 per cent versus 18.2 per cent, runs 2.5 times faster at the default setting, and cuts fabricated claims by more than five times. Partners reported runs 2 to 2.8 times faster using 12 to 23 per cent fewer tokens.
Argon still leads on Chartography at 71.6 against 66.6 per cent, and holds published scores on GraphWalks and SEC-bench Pro where Hydrogen has no figure. The practical split is that Hydrogen wins agentic and tool-use work while Argon keeps the heavier reasoning benchmarks.