FRI, NOVEMBER 13, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

Google’s Gemini 4 Hydrogen Outscores Its Own Flagship, Four Weeks After Argon Shipped

Gemini 4 Hydrogen launched 12 November at $1.50 input and $7.50 output per million tokens, flat across a 1M context. It scores 51 on the Artificial Analysis Intelligence Index against Gemini 4 Argon at 50, triples 3.8 Flash on Terminal-Bench 4.1, and runs 2.5x faster at default. Free users get it in Search and the Gemini app.

By AIToolsRecap November 13, 2026 6 min read 52 views
Home › Articles › News › Gemini › Gemini 4 Hydrogen: $1.50/M and It Beats Argon

● Price: $1.50 in / $7.50 out per million tokens, flat across the whole 1M context. Cache hits $0.075.
● The headline: it scores 51 on the Artificial Analysis Intelligence Index against Argon’s 50 — the smaller model beating the flagship four weeks after Argon launched.
● Speed: 2.5x faster than 3.8 Flash at default. Partners report 2x to 2.8x faster runs and 12–23% fewer tokens.
● Where: Gemini API and AI Studio on all tiers, Android Studio, the Gemini app, and Google Search globally — free users included.

Google announced Gemini 4 Hydrogen on 12 November, positioning it as the “daily driver for agentic workflows” rather than a deep reasoning model. It replaces 3.8 Flash and adds a fourth thinking level, xhigh, on top of low, medium and high.

The benchmark that matters most is Terminal-Bench 4.1, where Hydrogen scores 57% against 18.2% for 3.8 Flash — more than triple. Google also reports fabricated claims cut by more than five times against its predecessor. On agentic coding, SWE-sweep puts Hydrogen at 13.5% against Argon’s 4.3%.

Benchmark Hydrogen Gemini 4 Argon
AA Intelligence Index 51 (54 at xhigh) 50
SWE-sweep, agentic coding 13.5% 4.3%
Agents’ Last Exam v1.1 71.6% 69.0%
OSWorld-2.1, offline subset 77.6% 74.7%
Vending-Bench 2, balance $14,549 $13,718
Chartography 66.6% 71.6%

Argon still wins where you would expect a bigger model to: Chartography, GraphWalks at 84.2%, SEC-bench Pro at 83.4% and Terminal-Bench 4.0 at 57.1%, none of which Google published a Hydrogen figure for. The honest read is that Hydrogen wins the agentic and tool-use benchmarks and Argon keeps the heavier reasoning ones.

Elsewhere Hydrogen takes 58.8% on HLE-Diamond, 52.3% on the Vals Tax Agent Bench for first place overall, 55.8% on Vals Legal Research for a tie at the top, 53.4% on the Multimodal Knowledge Agent set, and fifth on Agent Arena. Artificial Analysis puts cost per task at $0.29 at low effort rising to $3.16 at xhigh.

Full comparison: Hydrogen vs Argon, benchmark by benchmark · Hydrogen vs Claude Fable 5.2 on price. Listing: Gemini 4 Hydrogen.

Sources

Tags
AI NewsGoogleGenerative AI2026AI agents

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →