Google has launched Gemini 4 Hydrogen globally — replacing Gemini 3.8 Flash as the default model for everyday AI tasks across Google Search, the Gemini app, and developer tools like Google Antigravity. It is not hardware. It is not a phone, tablet, or laptop feature. It is a new AI model, live now, free to all users in those services, with no subscription required.
No new device needed. No upgrade prompt. Just faster, cleaner, more visual answers, starting today. Hydrogen is the first Gemini 4 model built explicitly for speed and reliability on routine workloads: web search, email drafting, file summarisation, and light coding.
Google says it runs up to 2.5 times faster than 3.8 Flash while reducing factual errors and sticking more closely to user instructions on the first try, even when those instructions include tone, formatting, or context pulled from Gmail or Google Drive. It replaces Flash in the Gemini app’s model picker for general use, though 3.5 Flash-Lite remains available for ultra-fast, low-complexity replies.
In Search, AI Mode now runs by default, powered by Hydrogen. That means AI Overviews handle multi-part questions, typos, and follow-ups more coherently, and deliver richer outputs: comparison tables, custom tools, and broader source-link coverage across the web. Google calls this “the biggest model upgrade ever added to Search.” It also marks the formal end of the Flash branding.
From here on, Gemini models will be named after periodic table elements, Hydrogen, Carbon, Argon, Flash is gone. Hydrogen sits between Argon (the heavy-duty reasoning model) and Carbon (the architecture-and-planning specialist) in Google’s new tiered strategy. It does not replace either. Instead, it handles fast-turnaround tasks at lower token cost, freeing up heavier models for deeper work.
On the Artificial Analysis Intelligence Index, Hydrogen scores slightly higher than Gemini 4 Argon on accuracy, but finishes tasks faster. Developers get native support for a 1-million-token context window, plus access to a new Orchestrator tool inside Google Antigravity. Orchestrator dynamically routes parts of a coding task, like unit testing or boilerplate generation, to Hydrogen, while reserving Argon or Carbon for high-stakes design decisions.
API pricing starts at $1.50 per million input tokens and $7.50 per million output tokens via Google AI Studio and Gemini for Business. Safety filters have been updated too: fewer false refusals on legitimate science or medical queries, tighter control over automated tool use, and stricter adherence to user permissions.
No launch event. No press release headline about “revolutionising” anything. Just a quiet, global rollout, live as of 2026-11-12. For most users, that means typing a messy question into Google Search or asking Gemini to rewrite a paragraph will just work better, right now.
No opt-in. No toggle. No new app. Gemini 4 Hydrogen is live.
And it’s already running.
ADFiled by The AI Desk
Models, assistants and the companies and chips behind them, reported from what was released and what was claimed, with the difference kept clear.
More from this desk →
Be the first to comment
Join the argument. No password, just your email or a passkey.