Google launches Gemini 4 Hydrogen, bringing its Gemini 4 models to free users
Archived — this story has rotated out of today’s deck. It is kept here in full.
The gist
Google released Gemini 4 Hydrogen, priced at $1.50 per million input tokens and $7.50 per million output tokens for developers.
Hydrogen is the first Gemini 4 model available to free users, not just paying subscribers.
Background
Google DeepMind is Google's AI unit behind the Gemini 4 family of models, which so far includes Argon and Carbon. Hydrogen is a faster, cheaper 'workhorse' model in that family, and unlike Argon and Carbon, which are limited to paying users, it brings Gemini 4 to all tiers. The launch follows a year in which Google delayed and then canceled Gemini 3.5 Pro, and many considered the company to have fallen behind in the race for the most advanced AI models, known as the frontier.
How it unfolded
- Sep 2, 2026Google launched Gemini 3.8 Flash; Hydrogen is later billed as a major standalone upgrade for teams migrating from it.
- Sep 30, 2026Google announced Gemini 4 Argon, with model ID gemini-4-argon and an introductory price of $2 input and $10 output per million tokens.
- Nov 7, 2026A report said X users claimed Google was preparing a Flash model called Gemini 4 Hydrogen; no official Google source named the model.
- Nov 10, 2026Google announced Carbon availability; its post said Carbon uses more tokens than Argon to explore thoroughly.
- Nov 11, 2026Business Insider reported Google had been internally testing a model known internally as Hydrogen, with staff access in the internal development platform Jetski; Google declined to comment on the record.
- Nov 12, 2026Google announced Gemini 4 Hydrogen, rolling out to the Gemini app, Google Search, Google AI Studio, the Gemini API, Gemini for business, and paid Google AI Pro and Ultra subscribers in Google Antigravity.
- Nov 12, 2026Hydrogen was priced at $1.50 per million input tokens and $7.50 per million output tokens for developers; Google Cloud's public SKU catalogue listed the same rates, though the Gemini API and Vertex AI pricing pages had no Hydrogen entry when checked at 10:31 UTC.
- Nov 12, 2026Google said Hydrogen finishes tasks 2.5 times faster at its high default setting, more than triples Terminal-Bench 4.1 scores (57% vs. 18.2%), cuts fabricated claims by more than 5x, and surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level (51 vs. 50, reaching 54 at its highest xhigh setting).
Key terms
- tokens
- A machine learning model trained on huge amounts of text to generate, summarize, translate, and analyze language.
- agentic workflows
- AI programs that can make decisions, pursue goals, and take actions independently, often used for coding tasks.
Source: Wikipedia
Who’s saying what
- Official
- Google says Hydrogen brings 'cutting-edge intelligence to everyone — with speed and efficiency for everyday coding and agentic workflows,' and is a major standalone upgrade over Gemini 3.8 Flash.
- Expert
- Shattered.io said readers should 'treat every Hydrogen specification you read online as invented' until Google confirms or denies the model, in a report written before the launch.
- Analysts
- A Google employee told Business Insider that Hydrogen is akin to Google's previous Flash models, positioned as a workhorse faster and more cost-effective than models like Argon.
Still unverified
Before the launch, the existence of Gemini 4 Hydrogen, its Flash classification and the X origin of the claim rested on social posts and an OfficeChai write-up with no official Google source; Business Insider also reported it was not immediately clear whether Hydrogen, if released, would be a paid or free offering.