Google released Gemini 4 Hydrogen, a new AI model that replaces Gemini 3.8 Flash across its products. The rollout is immediate and unrestricted, unlike the staged, security-focused releases of earlier models Argon and Carbon.
Google claims Hydrogen delivers major improvements over Flash 3.8. In high thinking mode, Hydrogen runs 2.5 times faster than Flash 3.8. The model reduces fabricated claims by more than 5x according to Google's analysis. Hydrogen more than triples Flash 3.8's score on Terminal-Bench for code generation. In the Artificial Analysis Intelligence Index with xhigh mode, Hydrogen scored 54 versus 50 for Argon.
Hydrogen runs at four thinking levels: low, medium, high, and xhigh. Google positions it as ideal for developers building agentic coding tools, generating simpler code, and running tests without high token costs. Argon and Carbon remain Google's choice for long-horizon tasks and security-sensitive work.
Google is integrating Hydrogen into Google Search, calling it "the biggest update to its AI-powered search tools ever." The model should better handle long, multipart questions. AI Overviews and AI Mode responses will include more visual elements. Google claims Hydrogen's search fan-out links to more sources.
Access varies by product. Everyone gets Hydrogen in Search. In the Gemini app and AI Studio, AI Pro and Ultra subscribers have Hydrogen, while AI Plus and free users still use the old 3.5 Flash-Lite model. Developers can access Hydrogen via API at $1.50 per million input tokens and $7.50 per million output tokens, matching Flash 3.8 pricing. Carbon and Argon pricing rises to $4 and $20 per million tokens next month from current $2 and $10.