Google launches Gemini 4 Hydrogen with 1M-token context and Antigravity Orchestrator

Google has introduced Gemini 4 Hydrogen, a new model focused on faster responses, coding, everyday tasks, and agentic workflows at a lower cost than Gemini 4 Carbon and Argon. It supports a one-million-token context window, text, image, audio, and video inputs, and four reasoning levels: low, medium, high, and xhigh. On Terminal-Bench 4.1, Hydrogen scored 57%, compared to 18.2% for Gemini 3.8 Flash, while its Intelligence Index score reaches 54 at xhigh. It also outperforms GPT-6.1 Terra and Claude Sonnet 5.5 in several benchmarks, although Sonnet remains ahead in computer-use tasks.

Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, making it 25% cheaper than the promotional pricing of Carbon and Argon, but 50% more expensive than GPT-6.1 Terra. In the Gemini app, Google says the model better follows instructions related to tone, formatting, and style, while improving structured responses and tool selection. Free users can also access it through automatic model selection.

Google has also introduced Orchestrator in Antigravity, which switches between Carbon for complex tasks and Hydrogen for routine steps. According to Google, this reduces quota consumption by 24% and speeds up task completion by 19%. Hydrogen is now available through Google AI Studio, the Gemini API, Gemini for business, Antigravity, the Gemini app, and Google Search.