Overview
- Google has begun a global rollout of Gemini 4 Hydrogen across Search, the Gemini app, the Gemini API, Google AI Studio, Antigravity and paid Pro/Ultra tiers, with free-tier queries routed automatically.
- Hydrogen is pitched as a high-speed, workhorse model that Google says is about 2.5 times faster than Gemini 3.8 Flash and can handle multi-step, agentic workflows on its own.
- The model uses a 'fan-out' approach that breaks long questions into sub-queries so Search in AI Mode delivers richer, more visual answers such as images, comparison tables and structured results.
- Hydrogen offers four selectable 'thinking levels' (low, medium, high and extra-high) to trade off speed against depth of reasoning and output token use, and it can be paired with heavier models Carbon and Argon for harder problems.
- Adoption and real-world cost will depend on token pricing, gated access to security-focused variants and limited independent verification of Google’s benchmark and safety claims, which leaves questions about sustained multi-agent costs and dual-use risks.