Gemini 4 Hydrogen is GA at $1.50 / $7.50
As of 2026-11-13, Gemini 4 Hydrogen has shipped: Google marked gemini-4-hydrogen as generally available in the 2026-11-12 Gemini API release notes, at $1.50 per 1M input tokens and $7.50 per 1M output tokens, listed on both the Free Tier and the paid tier (the notes write Free of charge for the former, rate-limited). Below are the official specs, thinking levels, how it differs from Carbon and Argon, and how to call it.
Updated 2026-11-13
Claude, GPT and Chinese models on one key, billed per token; check /models for callable models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four key points
Official price per 1M tokens
The pricing page and the what's new page both list $1.50 input and $7.50 output, with thinking tokens billed at the output rate; cached input is $0.075 through implicit caching.
The official tier name, written Free of charge
The pricing page and the what's new page both list the model on this tier and write Free of charge, rate-limited; the Carbon and Argon sections say not available on the Free tier.
Context window, input and output shared
Input, thinking and output share one 1,048,576-token window; max output defaults to 65,536 and long decode goes up to 1,048,576.
The fourth thinking level
Four levels: low, medium, high (default) and xhigh; sending minimal returns an error. Carbon and Argon default to medium and cap at high.
What Gemini 4 Hydrogen is
The model page positions it as cutting-edge intelligence for everyone, aimed at everyday coding and agentic workflows, with an emphasis on speed and efficiency. Inputs are text, images, video, audio and PDF, and output is text; function calling, structured outputs, code execution, computer use, file search, Search and Maps grounding and URL context are all supported, while audio generation, image generation and the Live API are not. Caching is implicit only; explicit caching is not supported. The same release note says it is now the default model for the Antigravity agent in Gemini Managed Agents.
What the 2026-11-12 release note says
The 2026-11-12 entry of the Gemini API release notes marks gemini-4-hydrogen as GA and states: it is available on the Free Tier and the paid tier; paid tokens are billed at the standard prices on the official pricing page; it adds a fourth thinking level, xhigh; and it becomes the default model for the Antigravity agent in Gemini Managed Agents. Migration guidance points to Gemini 3.8 Flash. Before it, Carbon went GA on 2026-11-10 and Argon on 2026-10-13.
Official timeline
Gemini 4 Argon (gemini-4-argon) went GA; Google describes it as the efficient Gemini 4 frontier reasoning model.
Gemini 4 Carbon (gemini-4-carbon) went GA; Google calls it its most capable reasoning model, at an introductory $2.00 / $10.00 through 2026-11-30.
Gemini 4 Hydrogen (gemini-4-hydrogen) went GA, adding the fourth thinking level xhigh and becoming the default model for the Antigravity agent in Gemini Managed Agents.
What Google confirms and what it leaves open
Officially confirmed
All of this comes from Google pages fetched on 2026-11-13: (1) the model ID is gemini-4-hydrogen, the version is Stable and it went GA on 2026-11-12; (2) the unified context is 1,048,576 tokens, max output defaults to 65,536 and long decode can reach the window limit; (3) inputs are text, images, video, audio and PDF, output is text; (4) thinking levels are low, medium, high (default) and xhigh, and minimal returns an error; (5) the Free Tier is written Free of charge and rate-limited, while cached input on the paid tier is $0.075; (6) Batch and Flex are $0.75 / $3.75 and Priority is $2.70 / $13.50; (7) Batch, Flex and Priority are all supported; (8) the Antigravity agent uses it by default.
What Google has not published
Two things are not on the official pages: a knowledge cutoff date, and per-minute request and token numbers for the Free Tier (the pages only say rate-limited and point to the rate limits page). Both can only come from Google, and guessing from the model name has no basis.
Telling Hydrogen, Carbon and Argon apart
Gemini 4 Hydrogen (fast, and it has a Free Tier)
Google positions it as fast and efficient, for everyday coding, interactive web apps, tool-using agents and parallel subagents; the what's new page recommends it as the upgrade from 3.8 Flash and earlier Gemini 3 and 2.5 models. Of the three it is the only one listed on the Free Tier.
Gemini 4 Carbon and Argon (paid tier only)
Google calls Carbon its strongest frontier model, spending more thinking tokens per turn to test hypotheses and verify edge cases; Argon handles the majority of advanced long-horizon tasks with lower token use than Carbon. The pricing page says neither is available on the Free tier, and both carry the same introductory price of $2.00 / $10.00 through 2026-11-30, then $4.00 / $20.00 from 2026-12-01.
How to call it
Google gives two entry points that take the same model string: pick gemini-4-hydrogen in Google AI Studio, or set model to gemini-4-hydrogen on the Gemini API, which serves both GenerateContent and the Interactions API. The official REST example is POST https://generativelanguage.googleapis.com/v1beta/interactions with x-goog-api-key in the headers and model plus input in the body. Thinking defaults to high and synchronous output is capped at 64K; for longer output set max_output_tokens above 65,536 and turn on background, or use client-side continuation in the SDK. Coming from 3.8 Flash, change the model string first, then drop temperature, top_p and top_k and move from explicit to implicit caching.
On QCode
QCode does not offer the Gemini series, so gemini-4-hydrogen is not on /models. To call models per token, one QCode key covers Claude, GPT and Chinese models such as claude-opus-5-5, gpt-6.1-sol and glm-5.3, billed per token with per-model rates on /models. Three steps: create an account, top up, and for Claude models in Claude Code set ANTHROPIC_BASE_URL to https://api.qcode.cc/api. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan, add a booster pack when you need more quota.
Frequently asked questions
Is Gemini 4 Hydrogen out?
Yes. The 2026-11-12 Gemini API release notes mark gemini-4-hydrogen as generally available, and it now runs in Google AI Studio and the Gemini API.
How much does Gemini 4 Hydrogen cost?
Official pricing is $1.50 per 1M input tokens and $7.50 per 1M output tokens, with thinking tokens billed at the output rate; cached input is $0.075. Batch and Flex are $0.75 / $3.75 and Priority is $2.70 / $13.50. These are Google's official prices as of 2026-11-13.
Does Gemini 4 Hydrogen have a Free Tier?
Yes. The official pricing page and the what's new page list it on the Free Tier and write Free of charge, rate-limited; the Carbon and Argon sections say not available on the Free tier. The pricing page also marks the Free Tier as used to improve Google's products (Used to improve our products: Yes), while the paid tier says No.
How is Gemini 4 Hydrogen different from Carbon and Argon?
All three share the 1M unified context and long decode; they differ in positioning and thinking levels. Hydrogen is fast and efficient, defaults to high and adds xhigh; Google calls Carbon its strongest model and it spends more thinking tokens; Argon handles most long-horizon tasks with lower token use. Carbon and Argon default to medium and cap at high.
What is the model ID for gemini-4-hydrogen?
The ID is gemini-4-hydrogen and the version is Stable. The official REST example sets model to gemini-4-hydrogen, posts to https://generativelanguage.googleapis.com/v1beta/interactions and passes x-goog-api-key in the headers.
Can I use Gemini 4 Hydrogen on QCode?
No. As of 2026-11-13 QCode does not offer the Gemini series and gemini-4-hydrogen is not on /models; for similar work use the Claude, GPT or Chinese models on sale on /models.
Sources
Google: the Gemini API release notes (2026-11-12 entry), the Gemini 4 Hydrogen, Carbon and Argon model pages, the Gemini Developer API pricing page, What's new in Gemini 4 Hydrogen, Latest Gemini models, the models index, and the Thinking and Rate limits docs, all fetched and archived on 2026-11-13. The QCode side comes from qcode.cc /models and /pay-as-you-go, fetched the same day.
Ready to try Claude, GPT or a Chinese model?
Create a QCode key, pick a model ID from /models and pay per token; Pay-as-you-go Packs need no monthly plan and last 7 days.
Related reading
Pay-as-you-go Packs: no monthly plan, valid 7 days
Buy a one-time quota and use it within 7 days; per-model rates are on /models.
Gemini 4 Carbon API status
Carbon went GA on 2026-11-10, on the paid tier only; that page tracks its API status and dates.
Gemini 4 tracker
The Gemini 4 line: GA dates, specs and prices for Argon (2026-10-13) and Carbon (2026-11-10).
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Checked on 2026-11-13 against the Google Gemini API release notes, model pages, pricing page and what's new page fetched that day. All prices are Google's official prices and follow its official pages; QCode's available models and unit prices follow /models.