
Gemini 4 Hydrogen: what's new and who can use it
8 min read
Last updated November 13, 2026. How we cover news: we write from the official announcement and docs (linked below) and point you to the tools on Unloc. Not sponsored.

Short answer: Google announced Gemini 4 Hydrogen on November 12, 2026. It is a fast Gemini 4 model for everyday coding, agents and chat, and it replaces Gemini 3.8 Flash as Google's recommended everyday model. It is rolling out to everyone at once: Google Search, the Gemini app, the free and paid Gemini API, Google AI Studio and Gemini for business, with Google Antigravity for paid Google AI Pro and Ultra subscribers.
Sponsored
Your tool here: sponsor this list
Fixed $149/month on Draft with AI. A subscription, not Claim #1.
Pin this category · $149/moTwo days after Google made Gemini 4 Carbon, its strongest model, generally available, this is the release most people will actually touch. Hydrogen is now in Google Search for all users, in the Gemini app, and it is the model developers are told to move to from 3.8 Flash. Below we cover what Google announced, who gets it and where, the developer details that matter, and how it compares with the fast models from OpenAI and Anthropic. We worked from Google's launch posts, the Gemini API guide, the Gemini Apps help page and the model card, and we say where a figure is Google's own.
What Google announced
In "Introducing Gemini 4 Hydrogen", published on November 12, 2026, Google DeepMind's Koray Kavukcuoglu described Hydrogen as a model "with speed and efficiency for everyday coding and agentic workflows". Google positions Carbon and Argon for deep, long-running reasoning on the hardest problems, and Hydrogen for fast execution. The model card says Hydrogen is adapted from Gemini 4 Carbon.
The main points, all by Google's own account:
- Faster and more accurate than 3.8 Flash. Google says Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at its default setting, more than triples its Terminal-Bench 4.1 score (57% vs. 18.2%) and cuts fabricated claims by more than five times.
- Close to the bigger models. Google says Hydrogen scores 51 on the Artificial Analysis Intelligence Index at its default high thinking level, one point above Gemini 4 Argon, and 54 at its highest setting.
- Four thinking levels. Low, medium, high (the default) and a new xhigh level, so you only wait and pay for the reasoning a task needs.
- Long context. A 1 million token context window, with outputs that can run up to 1 million tokens.
- A new home in Search. Google calls it "the biggest model upgrade to Search yet".
Who gets Gemini 4 Hydrogen, and where
Unlike Carbon, which reached the Gemini app for Google AI Ultra subscribers only, Hydrogen is rolling out globally across Google's consumer, developer and business products from day one:
- Google Search: for all users, globally, per Google's Search post.
- Gemini app, free users: no model picker. Google's Gemini Apps help page says that with auto model selection on, most prompts use Flash-Lite for quick answers and prompts that need more reasoning may use Hydrogen. With auto selection off, free users get Flash-Lite.
- Gemini app, paid plans: Google AI Plus, Pro and Ultra subscribers can pick Hydrogen from the model drop-down.
- Gemini in Google Sheets: starting with Google AI Pro and Ultra subscribers, plus eligible business subscriptions.
- Developers: the Gemini API (free and paid tiers), Google AI Studio and Android Studio.
- Google Antigravity: paid Google AI Pro and Ultra subscribers, and business users.
- Gemini for business: through the API or the model drop-down.
- Other tools: Cursor's docs already list it as
gemini-4-hydrogenin Cursor.
For what each Google AI plan includes and costs today, check Google's live plans page.
What changes in Google Search
For most people, Search is where they will meet Hydrogen first. Google's Search post lists four changes:
- Better grasp of intent: longer, multi-part questions, typos and very specific requests are understood more precisely.
- Wider query fan-out: Search runs more parallel searches across the subtopics of a question and links to "a broader range of helpful links than before".
- More visual answers: more images, structured responses, comparison tables and custom tools.
- Richer answers in every supported AI Mode language.
If you publish content or run SEO, the fan-out change is the one to watch. Google says Search will now surface sites people "previously might not have discovered", but it has not published data on how this changes clicks, so treat any traffic effect as unknown until you see it in your own Search Console.
What's new in the Gemini app
In the Gemini app, Google says Hydrogen follows your tone, format and rules more closely, so drafts and rewrites need less back-and-forth. It also says the model calls the right tools more often and pulls context from your uploads, Gmail, Google Drive and connected apps such as Airtable and Dropbox. Google's help page lists low, medium and high thinking levels in the app. The new xhigh level is documented for the API, not the app.
For developers: API, Antigravity and Cursor
Google's Gemini API guide calls Hydrogen the "recommended upgrade from Gemini 3.8 Flash and earlier Gemini 3 and 2.5 models". The details that matter if you migrate:
- Model ID:
gemini-4-hydrogen, on both the free and paid tiers of the Gemini API. - Price: Google announced flat pricing across the full 1 million token window of $1.50 per million input tokens and $7.50 per million output tokens, with thinking tokens billed as output and cached input discounted through implicit caching. The free tier is rate-limited. Check Google's live pricing page before you budget.
- Defaults change: Hydrogen defaults to the high thinking level (3.8 Flash defaulted to medium), rejects the old "minimal" level, ignores temperature, top_p and top_k, and does not support explicit context caching.
- One shared context window: input, thinking and output share one window of about 1 million tokens, with a 64K default for a single response and longer outputs through background execution.
- Knowledge cutoff: March 2026, according to the model card. Google says the full safety report will follow "in the coming days".
In Google Antigravity, Hydrogen arrives with Orchestrator, a model option that hands work back and forth between Carbon and Hydrogen in one session: Carbon plans and unblocks, Hydrogen writes the edits and runs the tests. Google says Orchestrator reached near-Carbon success rates on Terminal-Bench while using 24% less quota and finishing 19% faster on average. The Gemini API's Antigravity agent now uses Hydrogen by default.
In Cursor, the model page recommends keeping Hydrogen on high for agentic coding and reserving xhigh for hard debugging, and points to Carbon for the hardest tasks.
How it compares with OpenAI and Anthropic's fast models
OpenAI. The closest match is GPT-6.1 Terra, which OpenAI announced on October 26, 2026 as a fast, lower-cost model for computer use, visual reasoning and everyday writing (see the OpenAI API changelog and our GPT-6.1 Terra explainer). Terra is in ChatGPT on paid plans and in the API. Hydrogen's broader reach is Search and the free Gemini app, where Google routes some prompts to it automatically.
Anthropic. In Anthropic's model lineup, Claude Sonnet 5.5 is described as "the best combination of speed and intelligence" and Claude Haiku 5.5 as the model for high-volume, fast tasks. Anthropic still points most work to Claude Opus 5.5, and Fable 5.2 for the hardest problems (see our Fable 5.2 guide). If you use Claude mainly for writing and long documents, Hydrogen is not a reason to switch on its own.
The benchmarks are Google's. Most of the figures above come from Google, measured on Google's chosen tests. Independent results on your own tasks matter more, and that is easy to check: Hydrogen is free to try in AI Studio and in the Gemini API's free tier.
Search tools. If you are weighing Google's AI answers against an AI search engine, our Perplexity vs Gemini comparison covers how Perplexity differs, and ChatGPT vs Gemini covers the wider assistant choice.
What it means if you are choosing tools
- Free Gemini users: you get Hydrogen when Gemini decides a prompt needs it, and in Search. You cannot pick it yourself without a paid plan.
- Paid Gemini users: try Hydrogen as your everyday model and keep Carbon or Argon (where your plan includes them) for long, hard tasks.
- Developers on 3.8 Flash: plan the migration. Check the changed defaults, remove sampling parameters and explicit caching, and test the medium and high thinking levels on your own evals.
- Coding teams: Hydrogen is a cheaper worker model to pair with Carbon in Antigravity or Cursor, not a replacement for your strongest model on hard architecture work.
- Publishers and marketers: watch your Search Console data in the coming weeks. Google says AI answers in Search will link to a wider range of sites, but the effect on clicks is not yet known.
For the bigger Gemini 4 model that launched this month, see Gemini 4 Carbon is now available: who gets it and what's new.
FAQ
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is a fast Gemini 4 model from Google, announced on November 12, 2026, for everyday coding, agents and chat. It is adapted from Gemini 4 Carbon, replaces Gemini 3.8 Flash as Google's recommended everyday model and is rolling out in Google Search for all users.
Is Gemini 4 Hydrogen free?
Partly. It is in Google Search for all users, and free Gemini app users get it automatically when auto model selection decides a prompt needs more reasoning. It is also on the free, rate-limited tier of the Gemini API and in Google AI Studio. Choosing it yourself in the Gemini app needs a Google AI Plus, Pro or Ultra plan.
Is Gemini 4 Hydrogen better than Gemini 4 Carbon?
No. Google says Carbon is its strongest model for long, hard problems. Hydrogen is faster and cheaper, and Google says it scores one point above Gemini 4 Argon on the Artificial Analysis Intelligence Index at its default setting. Google suggests pairing them, with Carbon planning and Hydrogen doing the edits and tests.
Does Gemini 4 Hydrogen replace Gemini 3.8 Flash?
Google's Gemini API guide calls Hydrogen the recommended upgrade from Gemini 3.8 Flash and earlier Gemini 3 and 2.5 models. Moving over means changing the model ID to gemini-4-hydrogen and checking a few changed defaults, such as the high default thinking level and no explicit context caching.
Can I use Gemini 4 Hydrogen in Cursor?
Yes. Cursor's docs list Gemini 4 Hydrogen with the model ID gemini-4-hydrogen and recommend the high effort level for agentic coding. It is also in Google Antigravity for paid Google AI Pro and Ultra subscribers.
Submit a tool if it belongs on the board. Listings are reviewed before they go live.
Share your learnings on socials
Other resources
View all
Best ChatGPT alternatives: free and paid AI assistants
The best ChatGPT alternative depends on the job: Claude is the closest all-round swap, Gemini fits Google users, Microsoft Copilot fits Microsoft users, Perplexity is built for cited research, Grok for live news and X, and Mistral Vibe and DeepSeek suit people who want a European option or open models.

Claude Sheets, Whiteboards and Meetings: what's new
Anthropic announced Claude Sheets, Claude Whiteboards and Claude Data on November 10, 2026, plus full apps inside Claude and Claude Meetings (coming soon). Here's what each does, who gets it, what Sheets can't do yet, and how it compares with ChatGPT, Gemini and Copilot.

Best Zapier alternatives: free, open source and AI options
The best Zapier alternative depends on why you are leaving: Make is the closest all-round swap with a visual builder, n8n and Activepieces let you self-host, Microsoft Power Automate suits Microsoft shops, Gumloop and Pipedream cover AI agents, IFTTT handles personal and smart home automations, and Workato is built for enterprise.