Google Puts Gemini 4 Hydrogen Into Search Worldwide at $1.50 per Million Tokens

Google DeepMind released Gemini 4 Hydrogen on November 12 and switched it on inside Google Search worldwide the same day, making it the third model in the Gemini 4 family in roughly a month. The company prices it at $1.50 per million input tokens and $7.50 per million output tokens — flat across a 1 million-token context window — undercutting Argon's introductory $2/$10 rates, which double to $4/$20 after November 30.

Koray Kavukcuoglu, SVP of Google DeepMind and Chief AI Architect, authored the announcement. Robby Stein, VP of Product for Google Search, published a companion post on the Search integration. Google positions Hydrogen as a faster, lower-cost daily driver for agentic workflows, while siblings Argon and Carbon handle deeper reasoning.

How fast did the Gemini 4 rollout happen?

The cadence is compressed and uneven. Argon launched October 13 across the Gemini API, AI Studio, the Gemini app and AI Mode in Search. Carbon followed with a restricted preview for cyber defenders in late October, then rolled out broadly on November 10 via the Gemini API, Google AI Studio, Google Antigravity and the Gemini app for AI Ultra subscribers. That's 28 days from Argon to Carbon's broad release, then just two days to Hydrogen.

Hydrogen is available immediately across all tiers of Google AI Studio, the Gemini API, Gemini for business, the Gemini app, Google Search, and paid Google AI Pro and Ultra subscribers in Google Antigravity. Free Gemini app users get Hydrogen through automatic model selection — Google's blog says it deploys "when it's the appropriate model for the prompt," but doesn't define that threshold.

What does Hydrogen change in Search?

Google Search now runs Hydrogen globally for all users. Stein's post says the model improves intent understanding for complex, multi-part queries, expands "query fan-out" to surface a wider range of links, and delivers more visual responses including structured tables and custom tools. It's the widest deployment yet for a Gemini 4 model.

What are the specs and benchmark claims?

The numbers Google and third-party documentation report:

• Pricing: $1.50 per million input tokens, $7.50 per million output tokens, flat across the 1 million-token context window.
• Output: up to 64,000 tokens in a single reply, per third-party documentation; Google's announcement cites an output context window of up to 1 million tokens.
• Thinking levels: low, medium, high (default) and xhigh.
• Benchmarks: 51 on the Artificial Analysis Intelligence Index at high thinking (Argon scored 50), rising to 54 at xhigh; 76.1% on AutomationBench and 58.8% on HLE-Diamond — beating Claude Sonnet 5.5 and GPT-6.1 Terra on those tests, Google claims.
• Speed: 243 tokens/second output, but time-to-first-token of 29.1 seconds on a log scale, well behind Claude Sonnet 5.5 (1.0s) and GPT-6.1 Terra (4.3s).
• Vs. older hardware: Google says Hydrogen finishes tasks 2.5x faster than Gemini 3.8 Flash at the high default setting and more than triples Terminal-Bench 4.1 scores, 57% vs. 18.2%.

One discrepancy stands out. Third-party API provider aihubmix.com lists a "Cache Read" price of $0.075 per million tokens for Hydrogen — a 95% discount on the input rate. Google's own announcement doesn't mention a cached-input discount.

Google also introduced Orchestrator in Google Antigravity, a system that dynamically switches between Carbon and Hydrogen within a single session. On Terminal-Bench 4.0, Google says Orchestrator hit near-Carbon success rates while using 24% less quota and completing 19% faster.

Partner testimonials from Cartwheel, LatchBio, Box, Pixelcut and Astrocade appear in the announcement but have not been independently verified.

What about safety?

The model card shows Hydrogen did not reach Critical Capability Levels for CBRN biological, cybersecurity or harmful manipulation uplift, though alert thresholds were reached for biological and cybersecurity, where Google says it continues to deploy mitigations. For the Loss of Control – Stealth and Situational Awareness TCL, Google says it is applying mitigations as if the threshold had been reached "as a conservative assumption." The company claims 99.9% chain-of-thought legibility with zero deceptive reasoning edits across 200,000+ audited turns, plus a 96.7% honesty rate on internal agent evaluations.

What happens next?

Two open questions frame the coming weeks: whether Carbon reaches general availability before Argon's price doubling on November 30, and how Google presents the three-model lineup to non-technical users. Source reporting also notes leadership has referenced Hydrogen alongside whispers of Krypton and Xenon, suggesting the periodic table isn't finished. Whether Hydrogen's faster-but-shallower profile improves Search quality at scale — or simply shifts the error distribution — will take real-world usage to judge.