Source time Edition date
Introducing Gemini 4 Hydrogen
推出 Gemini 4 Hydrogen
Google's Gemini 4 Hydrogen targets agentic workflows, but teams must weigh faster agent builds against deeper dependence on one vendor's stack.
Judgment
Engineering leads choosing Gemini 4 Hydrogen as their default execution model gain a flat $1.50/$7.50 per-million price and 24% lower Orchestrator quota burn, but accept tighter Google Antigravity lock-in and dependence on one vendor’s routing logic.
Basis
Google DeepMind SVP and Chief AI Architect said Hydrogen targets everyday coding and agentic workflows; Google positions it as a fast execution engine that can pair with Carbon and Argon. The mechanism is routing: Orchestrator in Google Antigravity switches models within a session, reaching near-Carbon Terminal-Bench 4.0 success while consuming 24% less quota and finishing 19% faster. Hydrogen is priced flat across its 1 million-token context at $1.50 per million input and $7.50 per million output, and it scores 51 versus Argon’s 50 on the Artificial Analysis Intelligence Index at default high thinking, 54 at xhigh. For teams replacing Gemini 3.8 Flash, Google claims 2.5x faster task completion and Terminal-Bench 4.1 at 57% versus 18.2%.
- Intelligence Index vs Argon51 vs. 50
- Input price per million tokens$1.50
- Output price per million tokens$7.50
- Quota use under Orchestrator24% less
Unclear The source does not disclose Orchestrator’s routing failure rates, per-model latency overhead, or whether quota savings hold outside Google Antigravity.
Next Check whether Antigravity’s Orchestrator exposes per-model routing logs and quota accounting in the next release, so teams can verify the 24% savings independently.