A coding assistant that writes a missing unit test has a different job from one tracing a timeout across three services. Google released Gemini 4 Hydrogen on Nov. 12 for everyday coding and tasks that use tools. Developers can try it in Google AI Studio or call it through the Gemini API. The model offers four levels of reasoning effort, with high as the default. Google’s launch · Developer announcement
Google lists API prices of $1.50 per million input tokens and $7.50 per million output tokens across the full million-token context window. Changing the reasoning level can still change how long a request takes and how much it costs. A call that spends longer working through a defect can use more tokens even when the price per token stays the same. Google’s pricing and reasoning settings
In Google Antigravity, a feature called Orchestrator can switch between models such as Hydrogen and Carbon during one session. Google reports that on Terminal-Bench 4.0 Orchestrator came close to Carbon’s task success rate while using 24% less quota and finishing 19% faster on average. Those are results from Google’s test of its routing feature. Google’s benchmark
Replay a routine change and a defect that sent someone through several files. Run them at two reasoning settings with the same tests and acceptance criteria. Count the retries and the reviewer’s corrections alongside the API bill. That shows which work benefits from spending longer on a request.