Gemini 4 Hydrogen Edges Argon, Opens to All

Google on Thursday, Nov. 12, released Gemini 4 Hydrogen, a faster and cheaper member of its Gemini 4 family that it says edges out its own frontier model, Gemini 4 Argon, on a widely watched independent index. The model went live the same day across the Gemini API, the Gemini app and Google Search, according to a blog post by Koray Kavukcuoglu, Google DeepMind's senior vice president and Google's chief AI architect.

Key takeaways

• Hydrogen scores 51 on the Artificial Analysis Intelligence Index at its default setting, one point above Argon, and 54 at a new "xhigh" thinking level, Google says.
• Paid API pricing is a flat $1.50 per million input tokens and $7.50 per million output tokens across a 1 million-token context window, with a free tier in Google AI Studio.
• It is the first Gemini 4 model open to free users, in Search and, via automatic model selection, in the Gemini app.

Gemini 4 Hydrogen price vs. Gemini 3.8 Flash and Argon

Google positions Hydrogen as the upgrade path from Gemini 3.8 Flash, its previous workhorse model. The pricing math depends on the date. Google's Gemini API pricing page, checked Nov. 13, lists 3.8 Flash at an introductory $0.75 input and $3.75 output through Dec. 31, rising to $1.50 and $7.50 on Jan. 1, 2027 — the same as Hydrogen.

Once Argon moves to its standard rate on Dec. 1, Hydrogen's input price will be 37.5% of Argon's. Unlike Argon and Carbon, Hydrogen is available on the free API tier, where Google may use submitted content to improve its products.

Gemini 4 Hydrogen benchmarks

Against 3.8 Flash, Google says Hydrogen finishes tasks 2.5 times faster at its default setting, scores 57% on Terminal-Bench 4.1 versus 18.2%, and makes more than five times fewer fabricated claims. Early partners saw it run 2 to 2.8 times faster while using 12% to 23% fewer tokens, the company said.

Google also reported 58.8% on HLE-Diamond, 52.3% on the Vals Tax Agent Bench and 53.4% on its new Multimodal Knowledge Agent benchmark. Several claims compare Hydrogen with unnamed "models in its class," and all figures are Google-published.

Orchestrator pairs Hydrogen with Carbon

Google also introduced Orchestrator in its Antigravity coding app, which switches between Carbon and Hydrogen within one session. On Terminal-Bench 4.0, it reached near-Carbon success rates while using 24% less quota on average and finishing 19% faster, Google said.

Why it matters

Hydrogen extends a rollout that began with Argon and continued with Carbon, which Kavukcuoglu had signaled would come early. A model that matches the flagship on a composite index at a lower price narrows the gap between "cheap" and "smart" tiers. On safety, Google said Hydrogen cuts agentic overreach by up to 81% and scored 96.7% on an internal agent honesty test.

What's next

Developers on 3.8 Flash have until Dec. 31 before its price doubles, a natural window to test a switch. Independent reproductions of Google's benchmark claims are the next thing to watch.

FAQ

Can I use Gemini 4 Hydrogen for free? Yes, within limits. It is on the free tier of the Gemini API in Google AI Studio and in Google Search for all users; free Gemini app users get it when automatic model selection picks it.

What are the thinking levels? Low, medium, high (the default) and a new xhigh tier. Higher levels reason longer and use more tokens.

What is the model ID? gemini-4-hydrogen.

References

• Introducing Gemini 4 Hydrogen — Google (Nov. 12, 2026)
• Gemini 4 Hydrogen: everyday coding and agentic workflows for developers — Google (Nov. 12, 2026)
• Gemini Developer API pricing — Google (accessed Nov. 13, 2026)