Technology

Google releases Gemini 4 Hydrogen, a fast AI model replacing Flash

Illustration: Tezkun / AI. Not a photograph of the event.

Google has introduced the Gemini 4 Hydrogen AI model, which replaces the Flash models in its lineup. According to the company, it is already available to users worldwide in the Gemini app and Google Search.

Hydrogen is designed for everyday work, programming, and tasks performed by AI agents. In the Gemini app, it has replaced Gemini 3.8 Flash, and it has become the default model on the Antigravity agent development platform. Developers can access it via Google AI Studio and the API, including for free. Google calls the release the largest update to the AI models integrated into Search.

According to the company, the new model is about 2.5 times faster than Gemini 3.8 Flash (2 to 2.8 times in individual tests) and hallucinates facts more than five times less often. In the Artificial Analysis Intelligence Index ranking, as 3DNews reports, Hydrogen scored 51 points at the standard high reasoning level compared to 50 for Gemini 4 Argon, and 54 at the XHigh level. In the Terminal-Bench 4.1 test, Hydrogen scored 57%, while Gemini 3.8 Flash scored 18.2%.

The model accepts text, images, video, audio, and PDFs. The context window is designed for 1 million tokens, and the response can reach up to 1 million tokens. There are four reasoning levels: xhigh has been added alongside low, medium, and high. The price on the paid tier is $1.50 per 1 million input tokens and $7.50 per output; batch processing halves the rates.

At the same time, Google introduced the Orchestrator tool for Antigravity, which automatically selects a model based on the nature of the task. The company opened access to the flagship Gemini 4 Carbon to paid subscribers on November 10.

What it means

Gemini 4 Hydrogen can be used in the Gemini app and website, in Google Search, as well as via Google AI Studio and the API, including free access. On the paid API tier, 1 million input tokens cost $1.50, and output costs $7.50; batch processing is half the price.

Published 13 November 2026, 17:32 · No updates · Russian original

Discussion
  1. Loading comments…
No links or insults. One comment per 20 seconds.

Comments on this story live under its post in our Telegram channel. Comment on Telegram