- Context
- 1Mtokens
- Released
- Today
Cost/1M tokens
Gemini 4 Hydrogen is a model from Google DeepMind. The launch announcement centres on agentic workflows, with Google DeepMind saying the model stands out for speed and efficiency. It is built for everyday coding and agent-driven tasks.
- Multimodal model
- Accepts images, audio, video, files
- 1M context
- Reasoning
- Tools
- Structured output
How it compares
Additional source evidence
- Source model
- google/gemini-4-hydrogen
- Source family
- gemini
- Last updated
- 2026-11-12
- Knowledge cutoff
- 2026-03
- Model context
- 1M
- Model output limit
- 66K
- Open weights
- No
- File attachments
- Yes
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- No
Published benchmarks
| Benchmark | Score | Evaluation details | Evidence |
|---|---|---|---|
| SWE-sweep | 13.5 · % | — | Source |
| Terminal-Bench | 57 · cheating-adjusted | 4.1 | Source |
| Agents' Last Exam | 71.6 · score | 1.1 | Source |
| AutomationBench-AA | 76.1 · % | — | Source |
| Vending-Bench | 14549 · bank balance (USD) | 2 | Source |
| OSWorld | 77.6 · offline subset | 2.1 | Source |
| MKA | 53.4 · Climb WSR accuracy | — | Source |
| Chartography | 66.6 · % | — | Source |
| Terminal-Bench-Science | 51.4 · % | 0.1 | Source |
| HLE-Diamond | 58.8 · % | — | Source |
Complete source record (JSON)
{
"id": "google/gemini-4-hydrogen",
"name": "Gemini 4 Hydrogen",
"description": "Google's Gemini 4 model for everyday coding and agentic workflows, with four thinking levels and a 1M-token context window",
"family": "gemini",
"last_updated": "2026-11-12",
"knowledge": "2026-03",
"attachment": true,
"reasoning": true,
"tool_call": true,
"temperature": false,
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"limit": {
"context": 1048576,
"output": 65536
},
"benchmarks": [
{
"name": "SWE-sweep",
"score": 13.5,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "%"
},
{
"name": "Terminal-Bench",
"score": 57,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "cheating-adjusted",
"version": "4.1"
},
{
"name": "Agents' Last Exam",
"score": 71.6,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "score",
"version": "1.1"
},
{
"name": "AutomationBench-AA",
"score": 76.1,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "%"
},
{
"name": "Vending-Bench",
"score": 14549,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "bank balance (USD)",
"version": "2"
},
{
"name": "OSWorld",
"score": 77.6,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "offline subset",
"version": "2.1"
},
{
"name": "MKA",
"score": 53.4,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "Climb WSR accuracy"
},
{
"name": "Chartography",
"score": 66.6,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "%"
},
{
"name": "Terminal-Bench-Science",
"score": 51.4,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "%",
"version": "0.1"
},
{
"name": "HLE-Diamond",
"score": 58.8,
"source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/",
"metric": "%"
}
],
"release_date": "2026-11-12",
"open_weights": false
}Text provenance
Gemini app/webGoogle SynthID text watermarking
Google DeepMind says text generated in the Gemini app and web experience is watermarked with SynthID by adjusting token probabilities.
Official explanationFrequently asked questions
API and pricing
Input costs $1.50 per 1M tokens and output costs $7.50 per 1M tokens. Across 7 paid providers, input prices range from $0.75 to $2.70 per 1M tokens.
Google AI Studio is the cheapest provider at the lowest blended price (3:1 input to output). It charges $0.75 per 1M input tokens and $3.75 per 1M output tokens.
Yes. Cached input costs $0.075 per 1M tokens, a 95% discount off the uncached input price of $1.50 per 1M tokens, and cache writes cost $0.0833 per 1M tokens.
A typical request of 10,000 input tokens and 1,000 output tokens costs $0.0113 at the cheapest provider, Google AI Studio.
Last day, 1 endpoint reported uptime data and it was at 99% uptime or better, with the best uptime at 100.0% by Google.
Privacy and data
Google is the provider offering zero data retention (ZDR eligible or 0-day retention).
Capabilities and specs
The context window is 1M tokens (1,048,576) and the maximum output is 944K tokens (943,718).
Yes. Input modalities are audio, file, image, text and video.
Yes, tool calling (function calling) is supported.
Yes, structured output (JSON schema or JSON mode) is supported.
Yes, reasoning is supported and always on. Effort levels are xhigh, high, medium and low, with medium as the default. All 7 tracked providers expose reasoning, and a reasoning_effort setting is offered by a provider.
Quality and benchmarks
Matched on model kind, blended price and release date, the alternatives are Fugu Max by Sakana AI, Qwen3.8 Max (0902) by Qwen and Grok 4.7 by xAI, each at a blended price of $3.00 per 1M tokens.
Fugu MaxGemini 4 Hydrogen vs Fugu MaxQwen3.8 Max (0902)Gemini 4 Hydrogen vs Qwen3.8 Max (0902)Grok 4.7Gemini 4 Hydrogen vs Grok 4.7
Kimi K3 is the highest-ranked open-weight model on Arena Text at rank 19. Gemini 4 Hydrogen itself is closed.
Versions and lineage
Both have a 1M token (1,048,576) context. Gemini 4 Hydrogen costs $1.50 input and $7.50 output per 1M tokens, released November 12, 2026, while Gemini 4 Argon costs $2.00 input and $10.00 output per 1M tokens, released September 30, 2026.
Licensing and access
No, Gemini 4 Hydrogen is closed.
It is available on Google Vertex AI. There is no tracked endpoint on Azure or Amazon Bedrock.
Basics
Gemini 4 Hydrogen was made by Google DeepMind and was first listed on OpenRouter on November 12, 2026.
No. It is closed and the listed API price is $1.50 input and $7.50 output per 1M tokens.
Generated from Vertical Tabs data, updated November 12, 2026.



Discussion
No comments yet
Be the first to share your thoughts.