Gemini 4 Hydrogen
- Provider
- Status
- Current
- Context
- 1,000,000 tok
- Price
- $1.5 / $7.5 /MTok
Gemini 4 Hydrogen is a Google model built for fast work in agents, released on 12 November 2026. Google calls it “a fast, disciplined execution model” and sets it beside Gemini 4 Carbon and Gemini 4 Argon, which it describes as its models for long, deep reasoning (Google). It is on the Gemini API, in the Gemini app, in Google Search and in Google’s business and coding tools.
Artificial Analysis scored it 54 on the Intelligence Index v5.0.1 at its highest thinking setting, xhigh, at launch. It costs $1.50 per million input tokens and $7.50 per million output tokens, the same rate across its whole 1 million-token context window.
Quick specs
| Gemini 4 Hydrogen | |
|---|---|
| Released | 12 November 2026 |
| Availability | Gemini API, Gemini app, Google Search, Gemini for business, Antigravity, Gemini in Sheets |
| Input price | $1.50 / MTok |
| Cached input | 95% off (Artificial Analysis) |
| Output price | $7.50 / MTok |
| Context window | 1,000,000 tokens, one price throughout |
| Max output | 1,000,000 tokens |
| Thinking levels | Low, medium, high (default), xhigh |
| Input | Text, images, video, audio |
| Knowledge cutoff | Not published |
| AA Intelligence Index | 54 at launch (v5.0.1, xhigh) |
| Cost per AA index task | $3.16 |
| API model ID | gemini-4-hydrogen (in Google’s AI Studio links) |
Who can use it
Everyone, from 12 November 2026. Google began rolling it out worldwide on launch day (Google):
- Developers: on the Gemini API in Google AI Studio, and in Android Studio.
- Gemini app: in the model menu. Free users get it through automatic model selection, when Google judges it the right model for a prompt.
- Google Search: for all users.
- Businesses: in Gemini for business, through the API or the model menu.
- Antigravity: for Google AI Pro and Ultra subscribers and for enterprises. Antigravity’s Orchestrator can switch between Carbon and Hydrogen within one session.
- Gemini in Sheets: for eligible business plans and, for consumers, starting with Google AI Pro and Ultra subscribers.
Pricing
Google charges $1.50 per million input tokens and $7.50 per million output tokens, flat across the full 1 million-token context window, with no higher rate for long prompts. Artificial Analysis lists cached input at 95% off. Google’s announcement gives no batch price.
That is the same token price Gemini 3.8 Flash moves to on 1 January 2027, when its introductory $0.75/$3.75 ends. It is below the $2/$10 introductory price of Gemini 4 Carbon and Gemini 4 Argon, which both rise to $4/$20 after 30 November 2026.
Benchmark performance
Independent
| At Hydrogen’s launch (12 Nov 2026) | Gemini 4 Hydrogen | GPT-6.1 Sol | GPT-6.1 Terra | Gemini 4 Argon |
|---|---|---|---|---|
| Intelligence Index v5.0.1 | 54 (xhigh) | 55 | 51 | 50 |
| Cost per index task | $3.16 | $1.27 | $0.96 | n/a |
| Price (in/out per MTok) | $1.50 / $7.50 | $2 / $10 | $1 / $5 | $2 / $10 intro |
Source: Artificial Analysis for Hydrogen. The other models’ figures are Artificial Analysis’s readings as carried on our ranking on the day Hydrogen launched; Sol’s and Terra’s costs per task were measured on v5.0.
Intelligence. At launch, 54 on the index at xhigh, which Artificial Analysis listed as 16th of 78 models. Google says it scores 51 at its default high setting.
Cost per task. $3.16 per completed index task. Its tokens cost less than GPT-6.1 Sol’s, but a completed task costs more than twice as much, because it uses far more of them: Artificial Analysis measured 304 million output tokens to run the index at xhigh, against a median of 72.8 million, and describes it as “very verbose”.
Speed. At xhigh, Artificial Analysis measured 188 output tokens a second, with 40.55 seconds to the first answer token and 13.7 minutes per index task.
Google’s own figures
| Benchmark | Gemini 4 Hydrogen | Notes |
|---|---|---|
| AA Intelligence Index | 51 at high (default) | At launch, per Google; Gemini 4 Argon 50, and 54 at xhigh |
| Terminal-Bench 4.1 | 57% | Gemini 3.8 Flash 18.2% |
| HLE-Diamond | 58.8% | |
| Vals Tax Agent Bench | 52.3% | First at launch, per Google |
| Vals Legal Research Bench | 55.8% | Tied first at launch, per Google |
| Agents’ Last Exam v1.1 | 71.6% |
Source: Google, vendor-reported. Google also says Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at the default high setting, and that partners measured it 2 to 2.8 times faster with 12% to 23% fewer tokens. It published no SWE-bench Pro or SWE-bench Verified figure.
Safety. On its own internal tests, Google says Hydrogen takes misaligned actions or shortcuts on tasks 12 times less often and is up to twice as unlikely to step outside the permissions it is given.
How it compares
vs Gemini 3.8 Flash
Google pitches Hydrogen as an upgrade for teams using Gemini 3.8 Flash. When Hydrogen launched, Gemini 3.8 Flash scored 31 on the v5.0.1 index, 23 points lower. Google says Hydrogen finishes tasks 2.5 times faster and more than triples Gemini 3.8 Flash’s Terminal-Bench 4.1 score. Gemini 3.8 Flash costs $0.75/$3.75 until 31 December 2026 and then $1.50/$7.50, Hydrogen’s price.
vs Gemini 4 Argon
When Hydrogen launched, Gemini 4 Argon scored 50 on the v5.0.1 index, four points under Hydrogen at xhigh. Google’s own comparison, at Hydrogen’s default high setting, is 51 to 50. Argon costs $2/$10 until 30 November 2026 and $4/$20 after.
vs GPT-6.1 Sol
When Hydrogen launched, GPT-6.1 Sol scored 55 on the v5.0.1 index, a point higher, at $2/$10 and $1.27 per completed task, against Hydrogen’s $3.16.
vs GPT-6.1 Terra
When Hydrogen launched, GPT-6.1 Terra, OpenAI’s low-cost model for agent work, scored 51 on the v5.0.1 index, three points lower, at $1/$5 and $0.96 per completed task.
Known limitations
Verbose. At xhigh it used 304 million output tokens to run Artificial Analysis’s index, more than four times the median, so a completed task costs more than its token price suggests.
Slow to start at xhigh. Artificial Analysis measured 40.55 seconds to the first answer token at that setting.
No SWE-bench figure. Google published neither SWE-bench Pro nor SWE-bench Verified.
Automatic on the free tier. Google says free Gemini app users get it through automatic model selection, when it is the right model for the prompt.
Frequently asked questions
When was Gemini 4 Hydrogen released?
On 12 November 2026. Google began rolling it out worldwide that day on the Gemini API, in the Gemini app, in Google Search and in its business and coding tools.
Can I use Gemini 4 Hydrogen for free?
Partly. Google says free Gemini app users get it through automatic model selection, when it is the right model for the prompt, and that it is in Google Search for all users. Antigravity access, and Sheets access for consumers, start with Google AI Pro and Ultra.
How much does Gemini 4 Hydrogen cost?
$1.50 per million input tokens and $7.50 per million output tokens on the Gemini API, the same across its full 1 million-token context window.
Is Gemini 4 Hydrogen better than Gemini 4 Argon?
On Artificial Analysis’s Intelligence Index at launch, it scored 54 at xhigh to Argon’s 50, and Google’s figure at the default setting is 51 to 50. It also has a lower token price than Argon. Google describes Argon and Carbon as its models for deep, long reasoning, and Hydrogen as its model for fast agent work.
How is Gemini 4 Hydrogen different from Gemini 3.8 Flash?
At Hydrogen’s launch, Artificial Analysis scored it 23 points higher on the v5.0.1 index (54 to 31), and Google says it finishes tasks 2.5 times faster. Its $1.50/$7.50 price is what Gemini 3.8 Flash costs from 1 January 2027, after its introductory price ends.