Google has launched Gemini 4 Hydrogen, a faster, lower-cost model for coding and AI agents, and it is available to everyone through the Gemini app and Search.
Developers can use Hydrogen at $1.50 per million input tokens. A token is a small chunk of text, roughly a word or part of one, and developers are billed by how many the model reads and writes. Google said Hydrogen nearly matches Argon, its flagship model. That is the company's own comparison, and it is the claim to check once outside testers get hold of the model.
The $1.50 figure covers input only. Output tokens are usually billed at a separate rate, and that rate was not part of the announcement as reported, so the real cost of running Hydrogen is still unclear.
The launch follows a familiar pattern. A lab builds a flagship to set the headline benchmark, then ships a cheaper, quicker sibling to handle the volume. Coding tools and agents are where that split matters most. An agent often makes dozens of model calls to finish one task, so a small difference in price or speed multiplies quickly. A single chat reply does not work that way.
Putting the model in Search and the Gemini app, rather than limiting it to paid developers, also suggests Google sees it as cheap enough to run at consumer scale.
The open question is what "nearly" means. A small gap on a general benchmark can grow into a large one on long, multi-step coding tasks, where one wrong step spoils the rest.