Google Launches Gemini4 Hydrogen for Fast Execution at Flat $1.5/$7.5 Pricing

AI寒武纪 · wechat · 2026-11-13

Google has released Gemini4 Hydrogen, a new model built for everyday coding and high-speed agentic execution, available to all users. On the ArtificialAnalysis Intelligence Index, its default high tier scores 51, edging out Google's own frontier model Gemini4 Argon at 50, while the xhigh tier reaches 54. Pricing is flat within a 1 million-token context: $1.50 per million input tokens and $7.50 per million output tokens.

Alongside the model, Google is rolling out Orchestrator in Google Antigravity, which can dynamically switch between models like Carbon and Hydrogen within a single session: on Terminal-Bench 4.0 it achieves task success rates close to Carbon while using 24% less quota on average and running 19% faster. Compared with Gemini 3.8 Flash, Hydrogen is 2.5x faster, lifts its Terminal-Bench 4.1 score from 18.2% to 57%, and cuts fabricated claims by more than 5x.

The model offers four thinking levels—low/medium/high/xhigh—letting users trade wait time and cost as needed. On the consumer side, the Gemini app and Google Search are getting upgrades too: stronger instruction following and tool calling, context integration with Gmail/Drive and third-party apps, and faster prototyping iterations, while Search gains improved intent understanding, visual answers, and query fan-out. The rollout covers AI Studio, the Gemini API, enterprise offerings, the Gemini app, Google Search, and Antigravity subscribers.

Original post →

Google Launches Gemini4 Hydrogen for Fast Execution at Flat $1.5/$7.5 Pricing

More from Models