IT Home, November 13: Google has officially launched the Gemini 4 Hydrogen model, designed for fast execution and efficient performance, targeting everyday coding and agent workflow scenarios, bringing the full capabilities of the Gemini 4 series to users at all levels.
$1.50 per million tokens (Note: Approximately RMB 10.0 at current exchange rates)
Output: $7.50 per million tokens (approximately RMB 50.1 at the current exchange rate)
This pricing covers the full 1 million token context window with no tiered pricing structure. In the Artificial Analysis intelligence index, Hydrogen defaults to a high-thinking tier score of 51, surpassing Gemini 4 Argon’s 50, and can reach up to 54 at the highest xhigh tier, while typically completing tasks faster than models in the same class.
Google has simultaneously launched the Orchestrator intelligent scheduling feature, integrated into the Google Antigravity platform, which automatically switches between models such as Carbon and Hydrogen within the same session. In the Terminal-Bench 4.0 tests, this scheduling solution achieved task success rates approaching that of Carbon, while reducing average quota consumption by 24% and improving completion speed by 19%.
Compared to the previous-generation Gemini 3.8 Flash, Hydrogen achieves 2.5x faster task completion at its default high-thinking mode, with its Terminal-Bench 4.1 score rising from 18.2% to 57%, and a reduction of over 5x in hallucination rate.
The model offers four levels of reasoning depth: low, medium, high (default), and xhigh, allowing flexible adjustment of reasoning depth, latency, and cost based on task requirements. It supports native inputs of up to 1 million tokens and outputs of up to 1 million tokens, enabling processing of multimodal content such as large PDFs, spreadsheets, charts, audio, and video.
In terms of security, Hydrogen reduces agent privilege escalation by up to 81%, service under-provisioning issues by 79%, and misalignment behaviors and task shortcutting in tool usage and coding scenarios by 12-fold; internal agent integrity evaluations score 96.7%, chain-of-thought retention reaches 99.9%, and resilience against indirect prompt injection on the Gray Swan IPI benchmark improves by more than 2.3 times.
Starting from the release date, Hydrogen is gradually rolling out globally: developers can access it via the Gemini API, AI Studio, and the paid tier of Android Studio; enterprises can deploy it through Gemini for Business, Google Antigravity, and Gemini within Sheets; general users can select the model from a dropdown in the Gemini app, and free users in automatic model selection mode will also receive optimized Hydrogen capabilities, while Google Search has also integrated this model to enhance the user experience.