Google has introduced Gemini 4 Hydrogen, priced at $1.50 per million input tokens and $7.50 per million output tokens. It is rolling out today in the Gemini app, Google Search, and the Gemini API, and it finishes tasks 2.5 times faster than Gemini 3.8 Flash at the default high setting.
At the default high level, Google puts its Artificial Analysis Intelligence Index score at 51, against 50 for Gemini 4 Argon, and at 54 on xhigh.
On Terminal-Bench 4.0, that pairing reached near-Carbon success while using 24 percent less quota and finishing 19 percent faster.
Tap a lens to see what this story means for you.
Liked this? The Brief brings you the whole day in tech, verified, every morning.
Two minutes, free forever. What's in The Brief →
See what’s happening right now
The Feed runs all day — short, verified briefs the moment they break.
Open the FeedFollow @thecircuitry_
Every story we publish, as it happens. No noise between.
Reader-supported
The Circuitry is a passion project I've always wanted to build, and I love the work behind it.
Running it costs real money. APIs, hosting, time. To keep improving the site and growing this into something useful for everyone, those costs have to be covered.
Any contribution is appreciated. If not, no pressure. Thanks for reading.
Google is rolling out Orchestrator in Antigravity for Google AI Ultra subscribers. It pairs Gemini 4 Carbon with Gemini 4 Hydrogen in one session, and on one test it used less quota and finished faster than Carbon alone.
Google has opened pre-orders for Intelligent Eyewear made with Warby Parker, Gentle Monster and Samsung, with prices from $395 and sales starting December 10. The Gemini glasses work with Pixel, Samsung and iPhone models, though some features need an Android phone at launch.
Intel is working with RadixArk to optimize SGLang, an open-source inference stack, for Intel CPUs, GPUs, and accelerators. The November 12 release does not include a speed figure.
United is taking steps to stop unauthorized third-party AI agents from using customer accounts. The airline says those tools can expose logins and can make purchases, account changes, or mileage redemptions the customer did not intend.
IBM has released two experimental Granite 4.1 preview models trained to read and write LLMON, a markup language for model context. The 3 billion and 8 billion previews are meant to separate instructions from data, and the page calls that a learned behavior, not a guarantee.
