Google has added a fast, cheaper model to the Gemini 4 family. Gemini 4 Hydrogen sits below Carbon and Argon, is aimed at everyday coding and agent work, and is priced at $1.50 per million input tokens and $7.50 per million output tokens across its full 1 million-token context window. For anyone paying for Gemini by the token, it is the model most likely to carry the bulk of routine work.
The pricing comparison is the first thing to notice. Hydrogen costs the same as Gemini 3.8 Flash will cost from January, when its introductory rate ends, and less than the $2 and $10 Carbon and Argon charge during their own introductory offers, which end on 30 November. Google says Hydrogen beats Argon on the Artificial Analysis Intelligence Index at its default setting, 51 to 50, and reaches 54 at the new xhigh thinking level. Against 3.8 Flash, the company claims tasks finish 2.5 times faster, Terminal-Bench 4.1 scores more than triple from 18.2% to 57%, and fabricated claims fall by more than five times.
The rest of the announcement is a long list of Google's own results. Hydrogen scores 58.8% on HLE-Diamond, 52.3% on the Vals Tax Agent Bench and 55.8% on the Vals Legal Research Bench, and ranks fifth on arena.ai's Agent Arena. Early partners report runs 2 to 2.8 times faster with 12% to 23% fewer tokens than 3.8 Flash. On safety, Google says misaligned actions and shortcuts in tool use fell twelvefold and that the model resists indirect prompt injection more than 2.3 times better on the Gray Swan benchmark.
Hydrogen also reaches more people than its siblings. It is rolling out in AI Studio and the Gemini API at all tiers, in Gemini for business, in Google Search, and in the Gemini app, where free users get it through automatic model selection. In Google Antigravity it is limited to paid AI Pro and Ultra subscribers, and it powers the new Orchestrator option, which hands work between Hydrogen and Carbon. For developers choosing between Hydrogen and a rival fast model such as Claude's or OpenAI's smaller tiers, flat pricing across a long context is a genuine advantage.
The caveats are familiar. Nearly every number comes from Google, and several benchmarks, including the new Multimodal Knowledge Agent test, were introduced or run by Google itself. Independent evaluations will take a few weeks to appear. The comparison that matters most for budgets, Hydrogen against Carbon once Carbon's price doubles, depends on how often Hydrogen is good enough on its own.