Google's Cheaper Gemini 4 Hydrogen Scores 51 on an Independent Index, Though at Its Top Setting It Still Ranks 16th
Google released Gemini 4 Hydrogen on November 12, a model it pitches as the fast execution engine for coding and agent work beside the larger Gemini 4 Carbon and Argon. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1 million-token context and is rolling out in the Gemini API, AI Studio, the Gemini app, Search and Antigravity. Artificial Analysis scores it 51 on its Intelligence Index at the default high setting and 54 at xhigh, where it ranks 16th of 78 models; most other benchmark figures are Google's own.
Google's newest Gemini 4 model is not its strongest. It is its pitch for the work that fills most agent sessions: writing code, running tests and checking changes after a bigger model has decided what to do. Gemini 4 Hydrogen, released on November 12, is priced well below Gemini 4 Carbon. Artificial Analysis scores it 51 at the default setting; Google says that is one point above the larger Gemini 4 Argon. That does not make it a frontier model: at its highest setting, the same index ranks it 16th.
What Google released#
Google says Hydrogen is built for "rapid execution across diverse agentic workflows", while Carbon and Argon handle "deep, long-horizon reasoning". It costs $1.50 per million input tokens and $7.50 per million output tokens, flat across a 1 million-token context window. It has four thinking levels: low, medium, high (the default) and a new xhigh.
The model is rolling out to all tiers of Google AI Studio and the Gemini API, Gemini for business, the Gemini app and Google Search, and to paid Google AI Pro and Ultra subscribers in Google Antigravity. Free Gemini app users get it through automatic model selection when Google's router picks it for a prompt.
Google also introduced Orchestrator in Antigravity, which switches between Carbon and Hydrogen within one session. On Terminal-Bench 4.0, Google says Orchestrator came close to Carbon's task success while using 24% less quota and finishing 19% faster on average.
Claims and checks#
Artificial Analysis, which runs its own evaluations, lists Hydrogen at 51 on its Intelligence Index at the high setting and 54 at xhigh, matching the figures Google cites; Google says Argon scores 50. At xhigh, Artificial Analysis ranks Hydrogen 16th of 78 models on intelligence and puts its cost at $3.16 per index task, falling to $1.87 at high and $0.29 at low, where it scores 27.
The rest of the comparison is Google's. It says Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at the default setting, scores 57% on Terminal-Bench 4.1 against 18.2% for 3.8 Flash, and cuts fabricated claims by more than five times. It also reports 58.8% on HLE-Diamond and first place on Vals' Tax Agent Bench at 52.3%, and says Hydrogen ranks fifth on arena.ai's Agent Arena. Google's safety claims, including a 96.7% rate on its internal agent-honesty evaluation, are internal results.
The arrival follows Google's Gemini Decisions API a day earlier, which also sells a narrow model as a component inside larger agent systems.
Limits#
- Speed, Terminal-Bench, HLE-Diamond, safety and partner figures are Google's own; we checked only the Artificial Analysis index scores, rank and cost per task.
- Artificial Analysis' cost per task depends on its benchmark workload and token use, not just list price; other workloads will cost more or less.
- The Agent Arena and Vals placements are as Google reports them; we did not open those boards.
- Free Gemini app users get Hydrogen only when automatic selection chooses it, so access there is not guaranteed for every prompt.