Day 7 receipt
Gemini 4 Hydrogen
Day 5 of 7Claims on the record
We check on 19 Nov 2026
- They promised
- A fast Gemini 4 model for everyday coding and agents that edges Argon on one index, live for everyone starting today.
The claims we’ll check
- “Hydrogen surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level (51 vs. 50”
- “finishing tasks 2.5 times faster at its high default setting”
- “Starting today, Hydrogen is rolling out globally across our developer, enterprise, and consumer surfaces”
- “free users will get access to Gemini 4 Hydrogen via auto model selection when it’s the appropriate model for the prompt”
Quoted word for word from Introducing Gemini 4 Hydrogen. Day 1: it is live across Google's products, but the benchmark and speed claims are Google's own and independent hands-on reports are only starting.
Full evidence 4 claims · 3 user reports · 4 sources
Google DeepMind's fast Gemini 4 model for everyday coding and agentic work, with four thinking levels, rolled out the same day to the API, the Gemini app and Google Search.
Launch post: Introducing Gemini 4 Hydrogen
Why this stamp
Day 1: it is live across Google's products, but the benchmark and speed claims are Google's own and independent hands-on reports are only starting.
Google says Hydrogen edges its larger Argon model on the Artificial Analysis index at the default thinking level, finishes tasks 2.5 times faster than Gemini 3.8 Flash, and is rolling out globally in the API, the Gemini app and Search. The Decoder, TechCrunch and The Verge repeat those claims. The early Hacker News thread is mixed: one user calls it a good sub-agent for code implementation, another reports worse tail failures in agentic work, and many complain about the model naming. Free Gemini app users only get it when auto model selection picks it. We re-stamp on day 7 (19 November) and day 14.
Launch dayWhat they claimed, quoted word for word · 4
“Hydrogen surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level (51 vs. 50”
Google's own figure, repeated by The Decoder. Not yet independently confirmed here, and some testers say the benchmark lead does not show up in agentic work.
“finishing tasks 2.5 times faster at its high default setting”
Google's comparison against its previous fast model. No outside speed test found yet.
“Starting today, Hydrogen is rolling out globally across our developer, enterprise, and consumer surfaces”
Google's post, TechCrunch and The Verge agree it is live in Search, the Gemini app and the API. A Hacker News user saw it appear on a business Workspace account the same day.
“free users will get access to Gemini 4 Hydrogen via auto model selection when it’s the appropriate model for the prompt”
Free-plan access is not guaranteed: it only applies when Gemini's auto selection picks Hydrogen. Nobody has measured how often that happens.
Day 7What real users report · 3
A commenter says it is a Flash-class model that works well as a sub-agent for code implementation, with a higher-level model writing the spec and doing review.
Reply in a 550-point launch thread · Open the source
A self-described Gemini fan finds early results with Hydrogen and Carbon disappointing: benchmark gains traded for worse rare failures in agentic work, such as overreach and overwork.
Comment in a 550-point launch thread · Open the source
A commenter says it already appeared on their paid Google Workspace account on launch day, much faster than past Google rollouts.
Comment in a 550-point launch thread · Open the source
OthersWhat outlets and testers say · 4
Reports Hydrogen as Google's fast model, 2.5 times faster than 3.8 Flash by Google's account, and the engine behind a Search upgrade Google calls its biggest yet.
“Google’s latest Search upgrade promises better answers to your most complicated questions” · Open the source
Covers the Search and AI Mode upgrade and the new xhigh thinking level, and says Google claims it outperforms some rivals in its class.
“Google’s new Gemini model is better at understanding complicated questions” · Open the source
Confirms the index scores Google cites, notes Hydrogen's top setting sits just behind Claude and OpenAI's best, and says the lineup is getting confusing.
“Google's Gemini 4 Hydrogen outscores the bigger Argon just four weeks after its launch” · Open the source
Summarises the launch, including its role in Search AI Mode, and repeats Google's claims.
“Google announces Gemini 4 Hydrogen with ‘significant leaps’” · Open the source