Google released Gemini 4 Hydrogen on November 12, 2026, and it is the first Gemini 4 model that does not cost anything to try. Hydrogen scores 51 on the Artificial Analysis Intelligence Index, one point above Gemini 4 Argon, at $1.50 per million input tokens and $7.50 per million output tokens. Google built it as the fast, everyday model of the Gemini 4 family, and it replaces Gemini 3.8 Flash as the model Google tells developers to move to.
Google's headline promises Gemini 4 for everyone, but who actually gets Hydrogen depends on where you look. Developers get a real free tier. Free Gemini app users only get Hydrogen when Google's Auto mode decides a prompt needs it. This guide covers what Hydrogen is, who gets it on each plan and what it costs against 3.8 Flash, Argon and Carbon. It also shows where its benchmarks hold up, and where Claude Sonnet 5.5 still wins.
The Key Takeaways
- Released November 12, 2026: Gemini 4 Hydrogen is Google's fast Gemini 4 model, adapted from Gemini 4 Carbon, and the recommended upgrade from Gemini 3.8 Flash.
- Free, with a catch: the Gemini API and Google AI Studio have a free tier for Hydrogen, a first for Gemini 4. In the Gemini app, free users get it only through Auto. Picking it yourself starts with Google AI Plus at $4.99 a month.
- Price: $1.50 / $7.50 per million tokens. That is double 3.8 Flash today, but 3.8 Flash rises to the same rate on January 1, 2027.
- Benchmarks: 51 on the Artificial Analysis Intelligence Index at default settings and 54 at xhigh, against 31 for 3.8 Flash. Artificial Analysis spent $1.87 per index task against $7.77 for 3.8 Flash.
- Where it loses: Claude Sonnet 5.5 beats it on Terminal-Bench 4.1 (63.9% vs 57.0%), OSWorld-2.1 and Terminal-Bench-Science in Google's own table.
What Is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is Google's fast, low-cost Gemini 4 model, built for everyday coding, agent work and chat rather than for the hardest reasoning problems. Google DeepMind's model card says Hydrogen "is adapted from Gemini 4 Carbon", the family's strongest model. Its model ID on the API is gemini-4-hydrogen, and it shipped as a stable release on day one, not a preview.
Where Hydrogen Sits in the Gemini 4 Family
Hydrogen is the third Gemini 4 model since September 30. Gemini 4 Argon went public on October 13. Gemini 4 Carbon reached the paid API and Google AI Ultra on November 10. Google calls Argon and Carbon its frontier models and says they "excel at deep, long-horizon reasoning". Hydrogen is the execution model: it writes the code, runs the tests and answers the everyday question quickly.
Is Hydrogen the Gemini 4 Flash?
Hydrogen is, in practice, the Gemini 4 Flash people have been searching for, though Google never used the Flash name for a Gemini 4 model. Google's developer guide does list Hydrogen as the "Recommended upgrade from Gemini 3.8 Flash and earlier Gemini 3 and 2.5 models". So Hydrogen fills the slot people expected a Gemini 4 Flash to fill. No Gemini 4 Flash-Lite exists yet. Google AI's launch post makes the same split between the models:
Today, we’re growing the Gemini 4 model series by introducing Gemini 4 Hydrogen.
— Google AI (@GoogleAI) November 12, 2026
While Gemini 4 Carbon tackles your hardest problems, Hydrogen brings cutting-edge intelligence to everyone — with speed and efficiency for everyday coding and agentic workflows, surpassing Argon on the Artificial Analysis Intelligence Index at its default thinking level, at 2.5x the speed of Gemini 3.8 Flash, with tunable thinking levels built to work alongside Carbon in @antigravity.
Watch how the 3D Medic Managed Agent app built in @GoogleAIStudio uses Gemini 4 Hydrogen to inspect your 3D files, catch errors, and patch them before export.
Who Can Use Gemini 4 Hydrogen?
Google AI Plus, Pro and Ultra subscribers can pick Hydrogen in the Gemini app, and free users get it only part of the time. Google's launch post lists eight surfaces for the November 12 global rollout. Consumers get it in the Gemini app, Google Search and Gemini in Google Sheets. Developers get the Gemini API, Google AI Studio, Android Studio and Google Antigravity, and companies get Gemini for business. The details differ a lot by surface.
In the Gemini App, Plan by Plan
Hydrogen appears in the model picker on Google AI Plus, Pro and Ultra. Google's Gemini models help page lists these models for chats on November 12:
| Plan | Price | Models you can pick in chats | Hydrogen |
|---|---|---|---|
| No plan (free) | $0 | Auto or Flash-Lite | Only when Auto routes a prompt to it |
| Google AI Plus | $4.99 / month | Flash-Lite, Hydrogen, Pro | Yes, in the picker |
| Google AI Pro | $19.99 / month | Flash-Lite, Hydrogen, Pro, Argon | Yes, in the picker |
| Google AI Ultra | $99.99 or $199.99 / month | Flash-Lite, Hydrogen, Argon, Carbon | Yes, with the highest limits |
Google AI Plus is the cheapest way to choose Hydrogen yourself. Google's plans page now puts "Includes Gemini 4 Hydrogen" on the Plus card, the same way it puts Argon on Pro and Carbon on Ultra. Gemini 3.8 Flash no longer appears in the app's model list on any plan. When a paid plan runs out of usage, Google says the chat "will continue with Flash-Lite".
Free Users Get Hydrogen Only Through Auto
Free Gemini users cannot select Hydrogen. Google's launch post says free users get it "via auto model selection when it’s the appropriate model for the prompt". The help page is more direct: "Most prompts will use Flash-Lite for quick answers. Prompts that require more reasoning may use Hydrogen." If you switch Auto off, Google says "all responses will use Flash-Lite".
A free user who wants Hydrogen should leave Auto on and ask harder questions. Even then, there is no guarantee which model answered. The help page explains how to check which model generated a response on the web. Our guide to what free Gemini includes covers the other free limits.
In Google Search
Hydrogen is "now available globally in Google Search for all users", according to Robby Stein, VP of Product for Google Search. He calls it "the biggest model upgrade to Search yet". Google says Hydrogen handles multi-part questions and typos better and widens the query fan-out, so answers link to more sites. It also returns more tables, images and custom tools in all supported AI Mode languages. Search is where every user gets Hydrogen without a plan.
For Developers: API, AI Studio and Antigravity
Hydrogen is on both the free and the paid tier of the Gemini API and Google AI Studio. That is new for Gemini 4: Google's pricing page says Carbon and Argon are "not available on the Free tier". The free tier is rate-limited, and Google's pricing table marks free-tier data as used to improve its products, while paid-tier data is not.
Google Antigravity, Google's agentic coding app, gets Hydrogen for paid Google AI Pro and Ultra subscribers. Antigravity also adds a model option called Orchestrator, which starts a task on Carbon for planning and hands the edits and test runs to Hydrogen inside one session. Google says Orchestrator reached near-Carbon success rates on Terminal-Bench while using 24% less quota and finishing 19% faster. Orchestrator is rolling out to Google AI Ultra subscribers only. Hydrogen is also now the default model behind the Antigravity agent in Gemini Managed Agents.
Want to compare Gemini with Claude or ChatGPT on the same prompt without three subscriptions? Fello AI puts Gemini, Claude and GPT in one Mac app, so the comparison is a tab, not a new account.
Gemini 4 Hydrogen Pricing
Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens on the paid Gemini API, according to Google's API pricing page. The rate is flat across its whole 1 million token window. Cached input costs $0.075 per million tokens, a 95% discount. Google only supports implicit caching for Hydrogen, which runs automatically on prompts of at least 4,096 tokens. Thinking tokens bill as output.
| Model | Input / 1M tokens | Output / 1M tokens | Free API tier |
|---|---|---|---|
| Gemini 4 Hydrogen | $1.50 | $7.50 | Yes |
| Gemini 3.8 Flash | $0.75 (to Dec 31), $1.50 from Jan 1, 2027 | $3.75 (to Dec 31), $7.50 from Jan 1, 2027 | Yes |
| Gemini 4 Argon | $2.00 (to Nov 30), $4.00 from Dec 1 | $10.00 (to Nov 30), $20.00 from Dec 1 | No |
| Gemini 4 Carbon | $2.00 (to Nov 30), $4.00 from Dec 1 | $10.00 (to Nov 30), $20.00 from Dec 1 | No |
Hydrogen costs twice as much per token as Gemini 3.8 Flash for now. On January 1, 2027, Google's own pricing page doubles 3.8 Flash to $1.50 and $7.50, so from that day the two cost the same per token and the upgrade is free. Against Argon and Carbon, Hydrogen is 25% cheaper per token until November 30. From December 1, when both frontier models move to $4 and $20, Hydrogen costs less than 40% of their rate. Our Gemini pricing guide covers the app plans and every other model.
The token price undersells the gap, because Hydrogen also uses fewer tokens. Artificial Analysis spent $1.87 per Intelligence Index task on Hydrogen at default settings, against $7.77 for Gemini 3.8 Flash and $9.95 for Argon, priced at Argon's $4 and $20 rate.
Hydrogen was cheaper to run on that test than the model it replaces, at twice the token price.
Gemini 4 Hydrogen Benchmarks
Hydrogen scores 51 on the Artificial Analysis Intelligence Index at its default high thinking level and 54 at xhigh. That puts it level with GPT-6.1 Terra, one point above Argon, and below Carbon's 60 and Claude Sonnet 5.5's 55. Google DeepMind's launch post sums up the claim:
Meet Gemini 4 Hydrogen: built for rapid coding, multimodal synthesis, and everyday workflows.
— Google DeepMind (@GoogleDeepMind) November 12, 2026
It finishes tasks 2.5x faster than 3.8 Flash and surpasses Argon on the @ArtificialAnlys Intelligence Index.
Artificial Analysis Scores
Artificial Analysis lists Hydrogen's scores on its Hydrogen model page. When we checked on November 12, the independent benchmarker ranked these models as follows:
| Model (setting) | Intelligence Index | Output speed | Cost per index task |
|---|---|---|---|
| Gemini 4 Carbon (high) | 60 | 76 tokens/s | $7.76 |
| Claude Sonnet 5.5 (max) | 55 | 144 tokens/s | $10.02 |
| Hydrogen (xhigh) | 54 | 198 tokens/s | $3.16 |
| Hydrogen (high, default) | 51 | 236 tokens/s | $1.87 |
| GPT-6.1 Terra (max) | 51 | 70 tokens/s | $0.96 |
| Gemini 4 Argon (high) | 50 | 64 tokens/s | $9.95 |
| Gemini 3.8 Flash (high) | 31 | 111 tokens/s | $7.77 |
Hydrogen is the fastest model in that table by output speed, and its default setting jumps 20 index points over 3.8 Flash. GPT-6.1 Terra matches its score for about half the cost per task. Artificial Analysis also rates the xhigh setting as very verbose, which is where the extra three points get expensive.
Google's Table vs Claude Sonnet 5.5 and GPT-6.1 Terra
Google's own benchmark table puts Hydrogen ahead of Claude Sonnet 5.5 and GPT-6.1 Terra on seven of ten tests, and behind Sonnet 5.5 on the other three. Google ran Hydrogen at its highest thinking setting unless a test notes otherwise. The rival figures mix vendor-reported numbers, public leaderboards including Artificial Analysis, and runs Google did itself. Sonnet 5.5's OSWorld-2.1 and Agents' Last Exam scores are Google's own runs, according to Google's evaluation methodology.
| Benchmark | Hydrogen | Claude Sonnet 5.5 | GPT-6.1 Terra |
|---|---|---|---|
| SWE-sweep | 13.5% | 3.1% | 5.8% |
| Terminal-Bench 4.1 | 57.0% | 63.9% | not reported |
| Agents' Last Exam v1.1 | 71.6% | 60.4% | 69.8% |
| AutomationBench-AA | 76.1% | 71.8% | 60.2% |
| Vending-Bench 2 | $14,549 | $11,165 | $9,107 |
| OSWorld-2.1 | 77.6% | 83.9% | 74.2% |
| MKA | 53.4% | 52.0% | 43.8% |
| Chartography | 66.6% | 61.6% | 63.2% |
| Terminal-Bench-Science 0.1 | 51.4% | 59.9% | 58.3% |
| HLE-Diamond | 58.8% | 53.3% | 47.3% |
Hydrogen's wins cluster in knowledge work and multimodal reading: agent exams, spreadsheet automation, charts and visually complex files. Claude Sonnet 5.5 keeps the terminal coding, computer use and science rows. For a coding-heavy team already on Sonnet, this table is not a reason to switch. Our Claude Sonnet 5.5 guide has Anthropic's side.
Against Gemini 3.8 Flash
Google says Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at its default setting and more than triples its Terminal-Bench 4.1 score, 57% against 18.2%. Google also claims Hydrogen makes more than five times fewer fabricated claims, because it flags missing context instead of guessing. In partner tests on graph, SQL and data-heavy work, Google reports Hydrogen ran 2 to 2.8 times faster with 12% to 23% fewer tokens.
These are Google's numbers. The Artificial Analysis gap of 51 against 31 is the independent one.
Gemini 4 Hydrogen vs Argon vs Carbon: Which One to Use
Use Hydrogen by default, and step up to Argon or Carbon only for long, hard reasoning jobs. That is Google's own advice. Its developer guide says to start with the high setting and keep xhigh for multi-step problems that high cannot solve. It points to Carbon for "hard architecture, debugging, science, and security audits".
Hydrogen is the right pick for everyday chat, writing, research, spreadsheets, PDFs and most coding. It is the fastest and cheapest of the three, and it is on every Google AI plan.
Gemini 4 Argon is Google AI Pro's frontier model. Hydrogen now matches its index score at a fraction of its API price, so Argon's case rests on deeper reasoning per turn, not on the headline number. Gemini 4 Carbon is the strongest Gemini at 60 on the index, kept for the hardest problems. In the app, Carbon needs Google AI Ultra.
Thinking level matters as much as the model. Hydrogen has four thinking levels, low, medium, high and a new xhigh, where Carbon and Argon default to medium and stop at high. Google warns that xhigh "can cause overthinking on simpler prompts". In the Gemini app, the help page lists only low, medium and high for every plan, so xhigh is for the API and Antigravity for now.
What the Hydrogen Model Card Adds
Hydrogen's model card shows one real regression: unjustified refusals rose by 5.57 percentage points against Gemini 3.8 Flash, "particularly for multimodal prompts". That means Hydrogen refuses more harmless requests than the model it replaces. Image-to-text safety improved by 7 points in the same table.
Hydrogen has one 1,048,576-token window shared by input, thinking and output. Replies default to 65,536 tokens and can run to 1 million with background long decode, where 3.8 Flash kept separate 1M input and 64K output budgets. It reads text, images, audio, video and PDFs, and it writes only text, so it does not generate images or audio. Its knowledge cutoff is March 2026.
On frontier risk, Google found no material capability increase over Argon. Targeted tests put Hydrogen below the critical capability levels for biological and cyber risk, though it reaches the alert thresholds in both. Google says the full Safety and Security Report is coming soon.
The Verdict on Gemini 4 Hydrogen
Hydrogen is the most useful Gemini 4 model for most people, because it is the one they can actually reach. For the price of Google AI Plus, it gives Argon-level index scores at more than twice the speed of 3.8 Flash. For developers, it is the first Gemini 4 model on the free API tier, and it costs less per task than 3.8 Flash despite the higher token price. The two caveats are real. Free app users get Hydrogen only when Auto allows it, and Claude Sonnet 5.5 still leads on terminal coding and computer use.
If you were on 3.8 Flash, switch now. Waiting for its January price rise saves less than it looks.
Frequently Asked Questions
Is Gemini 4 Hydrogen free?
Partly. On the Gemini API and in Google AI Studio, Hydrogen has a free, rate-limited tier, the first Gemini 4 model to get one. In the Gemini app, free users get it only through Auto model selection, when Google decides a prompt needs more reasoning; most prompts still go to Flash-Lite. Choosing Hydrogen yourself starts with Google AI Plus at $4.99 a month.
When was Gemini 4 Hydrogen released?
Google released Hydrogen on November 12, 2026, as a stable model on the Gemini API. The same day it began a global rollout across the Gemini app, Google Search, Google AI Studio, Gemini for business and Google Antigravity. It came two days after Gemini 4 Carbon reached the paid API.
Is Gemini 4 Hydrogen the same as Gemini 4 Flash?
In practice, yes. Google never released a model called Gemini 4 Flash. Hydrogen is the fast, low-cost Gemini 4 model, and Google lists it as the recommended upgrade from Gemini 3.8 Flash. No Gemini 4 Flash-Lite has been released.
How much does the Hydrogen API cost?
Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, flat across the 1 million token window, with cached input at $0.075 per million tokens. That is double Gemini 3.8 Flash's current rate, until 3.8 Flash rises to the same price on January 1, 2027.
Is Gemini 4 Hydrogen better than Gemini 4 Argon?
On the Artificial Analysis Intelligence Index, slightly: Hydrogen scores 51 at default settings against Argon's 50, and it produces output several times faster at a lower price. Google still positions Argon as the deeper reasoner for long, complex tasks, and Argon stays exclusive to Google AI Pro and Ultra in the Gemini app.
The featured image for this article was created with the help of AI.
