# Google releases Gemini 4 Hydrogen at $1.50 per million input tokens

> The new Gemini 4 model is built for everyday coding and agents, runs in Search and the Gemini app, and Google says it edges out Gemini 4 Argon on a key index.

URL: https://claudegpt.si/news/gemini-4-hydrogen-fast-model-pricing  
Published: 2026-11-14T18:55:43.088Z  
Author: ClaudeGPT Si Newsroom  
Section: AI Models  
Topics: Google, Gemini, Gemini 4 Hydrogen, Language models, Antigravity  
Publisher: ClaudeGPT Si

## Key takeaways

- Google released Gemini 4 Hydrogen on November 12, 2026 at $1.50 per million input tokens and $7.50 per million output tokens.
- Google says it scores 51 on the Artificial Analysis Intelligence Index at default settings, against 50 for Gemini 4 Argon, and runs 2.5 times faster than Gemini 3.8 Flash.
- It is in Google Search for everyone, in the Gemini app, and in the Gemini API for all tiers.

Google released Gemini 4 Hydrogen on November 12, 2026, a fast, cheaper member of the Gemini 4 family built for everyday coding and agent work. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1 million-token context window, and Google says it scores slightly higher than Gemini 4 Argon on a widely used intelligence index while finishing tasks 2.5 times faster than Gemini 3.8 Flash, according to [Google's announcement](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/).

Hydrogen fills the gap below the two big Gemini 4 models. Carbon and Argon are meant for long, hard reasoning; Hydrogen is meant to do the bulk of the work fast. It is also the model most people will meet first: Google put it into Google Search for all users and into the Gemini app on day one.

## Gemini 4 Hydrogen price and limits

| | Gemini 4 Hydrogen |
| --- | --- |
| Input | $1.50 per million tokens |
| Output | $7.50 per million tokens |
| Context window | 1 million tokens in, up to 1 million out |
| Long-context surcharge | None; the price is flat across the window |
| Thinking levels | Low, medium, high (default), xhigh |

The flat price matters for anyone feeding whole repositories or long documents into a model. Many APIs charge more past a certain prompt size; Google says Hydrogen doesn't. The new xhigh level lets it think longer when a task needs it.

## What Google says it scores

These are Google's own numbers:

- **Artificial Analysis Intelligence Index**, a composite of several benchmarks: 51 at the default high setting and 54 at xhigh, against 50 for Gemini 4 Argon.
- **Terminal-Bench 4.1**, which tests agents working in a command line: 57%, against 18.2% for Gemini 3.8 Flash.
- **HLE-Diamond**, a set of hard expert questions: 58.8%.
- **Vals Tax Agent Bench**: first overall at 52.3%.

Google also says Hydrogen makes more than five times fewer fabricated claims than 3.8 Flash and is up to twice as unlikely to bypass permission limits when a task is blocked. Early partners saw runs 2 to 2.8 times faster and 12% to 23% fewer tokens than with 3.8 Flash on data-heavy enterprise tasks, according to Google.

Independent numbers are starting to appear. [Artificial Analysis](https://artificialanalysis.ai/) added Hydrogen on November 12 and, in its capability comparison by field, puts Hydrogen at xhigh one to seven points behind Gemini 4 Carbon and Claude Opus 5.5 in finance, legal and engineering tasks, and roughly level with GPT-6.1 Sol. That fits Google's pitch: close to the top models, for much less.

## Where you can use it

- **Free users**: in Google Search, and in the Gemini app when its automatic model picker decides Hydrogen fits the prompt.
- **Gemini app**: in the model menu.
- **Google AI Pro and Ultra subscribers**: also in Google Antigravity, Gemini in Chrome and Gemini in Google Sheets.
- **Developers**: all tiers of the Gemini API in Google AI Studio, plus Android Studio.
- **Businesses**: Gemini for business, Antigravity, and Sheets for eligible plans.

Google says the rollout is global.

## The Orchestrator: Carbon plans, Hydrogen works

Alongside the model, Google added an Orchestrator to Antigravity, its coding agent app, that switches between Carbon and Hydrogen in one session. Carbon coordinates, and Hydrogen runs the worker and checker agents. On Terminal-Bench 4.0, Google says the pairing came close to Carbon's own success rate while using about 24% less quota and finishing about 19% faster.

That is the real bet here. Agent work burns tokens, and most steps in a long task are routine. A model that is good enough for those steps at a fraction of the price changes what an agent costs to run. For more on the larger models, see our [Gemini 4 Carbon vs Argon comparison](https://claudegpt.si/news/gemini-4-carbon-vs-argon).

## What to watch

The benchmark claims are Google's. Watch the full Artificial Analysis listing and the coding leaderboards over the next week for speed and accuracy measured by others, and watch whether OpenAI and Anthropic answer with price cuts on their smaller models.

## Sources

- [Introducing Gemini 4 Hydrogen](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/), Google
- [Gemini 4 Hydrogen: everyday coding and agentic workflows for developers](https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-4-hydrogen/), Google
- [Google Search gets a major upgrade with Gemini 4 Hydrogen](https://blog.google/products-and-platforms/products/search/gemini-4-hydrogen-search-upgrade/), Google
- [Artificial Analysis](https://artificialanalysis.ai/), Artificial Analysis
