Gemini 4 Hydrogen Edges a Frontier Sibling at $1.50 per Million Input Tokens

Google’s fast, low-cost Gemini 4 model now outscores one of its frontier siblings on a broad intelligence benchmark. Gemini 4 Hydrogen, introduced by Google DeepMind on November 12, scores 51 on the Artificial Analysis Intelligence Index at its default high thinking level — edging Gemini 4 Argon’s 50 — and 54 at a new xhigh setting, one of four thinking levels. Pricing is flat across the full 1 million-token context at $1.50 per million input tokens and $7.50 per million output.

Against Gemini 3.8 Flash, Google says Hydrogen finishes tasks 2.5 times faster, lifts Terminal-Bench 4.1 scores from 18.2% to 57% and cuts fabricated claims more than fivefold. Partners report 2x to 2.8x speedups with 12% to 23% fewer tokens. These are vendor and partner figures; run your own evals before moving a workload.

The more useful detail for builders is the new Orchestrator in Google Antigravity, which switches between Gemini 4 Carbon and Hydrogen inside one session. Google reports near-Carbon success on Terminal-Bench 4.0 with 24% less quota and 19% faster completion. Carbon itself became broadly available on November 10 at $2 per million input tokens and $10 per million output, and Google notes it uses more tokens than Argon. At list price Hydrogen costs 75% of Carbon per token, so the routing case rests less on that gap than on Hydrogen handling the steps that never needed Carbon.

The takeaway for teams shipping agents is architectural. A tier label no longer tells you capability — a “fast” model can beat a “frontier” one on a given benchmark — so the planner-and-executor split is now worth testing on your own tasks. Hydrogen also rolled into Google Search globally the same day, where Google says its reasoning strengthens the query fan-out technique that links users to a broader range of sites. The same model that runs your agents now shapes which pages get cited in AI answers.