Gemini 4 Argon
Google's most capable Gemini model is built for hard coding, enterprise knowledge work and cyber defense. Here is who can use it today, when it opens up, what it costs and what it can and cannot make.
Getting access
How to get Gemini 4 Argon
What Google has said about access so far.
Today: trusted programs only
Cyber defenders in the Fairwind Program and trusted testers have access now.
Next: paid API and AI Ultra
Google says paid Gemini API customers and Google AI Ultra subscribers come next. No date has been announced.
Need images now?
Argon does not generate images. LazyKiwi's image generator runs Nano Banana, Google's image model, in the browser.
What it is
What Gemini 4 Argon is built for
Google DeepMind announced Gemini 4 Argon on September 30, 2026 as a frontier model for real-world coding, enterprise knowledge work and cyber defense.
Release date and access
Limited preview, wider release "soon"
Argon is first available to trusted cyber defenders through Google's Fairwind Program and to trusted testers. Google says it will then reach paid API customers and Google AI Ultra subscribers, but has not given a date. It is not listed in the Gemini API, Vertex AI or the Gemini app yet.
- Announced September 30, 2026
- Trusted defenders and testers first
- Next: paid API and Google AI Ultra, no date
Long output
Up to 1M output tokens
Google says Argon can write up to 1M tokens in one response, up from 64K, and DeepMind lists long-context support of up to 1M tokens.
- 1M output tokens
- Up to 1M context
- Long-video understanding (LVBench 91.7%)
Images and video
It reads images; it does not make them
Argon understands images and long video, but Google's announcement does not describe image or video generation. At Google those come from separate models: Nano Banana for images and Veo for video.
- Input: text, images, video
- Image generation: Nano Banana
- Video generation: Veo
Price
Launch price for the API
Google lists an introductory price of $2 per million input tokens and $10 per million output tokens, with 95% off cached input, rising to $4 and $20 after the launch period.
- $2 / $10 per 1M tokens at launch
- $4 / $20 afterwards
- Launch-period length not stated
Specs
Gemini 4 Argon specs at a glance
From Google's announcement and DeepMind's model page.
- Developer
- Google DeepMind
- Announced
- September 30, 2026
- Status
- Limited preview (trusted defenders and testers)
- Built for
- Software engineering, enterprise knowledge work (legal, finance), cyber defense
- Inputs
- Text, images, video
- Outputs
- Text and code; no image or video generation described
- Output length
- Up to 1M tokens
- API price
- $2 input / $10 output per 1M tokens at launch; $4 / $20 later
- In LazyKiwi
- Not available; for Google images use Nano Banana
Benchmarks
Gemini 4 Argon vs GPT-6 Astra vs Claude Opus 5.5
Scores from the comparison table on DeepMind's Gemini page (highest thinking setting, pass@1). Each lab picks its own benchmarks; read them as Google's view.
| Benchmark | Gemini 4 Argon | GPT-6 Astra | Claude Opus 5.5 |
|---|---|---|---|
| Vals Index | 68.9 | 63.1 | 67.0 |
| DeepSWE v1.1 | 77.9 | 74.1 | 74.2 |
| FrontierSWE v2 | 55.0 | 65.5 | 62.3 |
| Terminal-Bench 4.0 | 57.4 | 58.2 | 66.4 |
| OSWorld-2.0 | 69.2 | 72.6 | — |
| LVBench | 91.7 | 87.5 | 83.7 |
FAQ
Gemini 4 Argon FAQ
What is Gemini 4 Argon?
Google DeepMind's frontier model announced on September 30, 2026, aimed at real-world coding, enterprise knowledge work such as legal and finance, and cyber defense. "Gemini Argon" and "Gemini Argon 4" refer to the same model.
What is the Gemini 4 Argon release date?
It was announced on September 30, 2026 and released first to trusted cyber defenders and testers. Google says paid API customers and Google AI Ultra subscribers come next but has not published a date.
Can I use Gemini 4 Argon in the Gemini app?
Not yet. As of mid-October 2026 it is not listed in the Gemini app, the Gemini API model list or Vertex AI.
Can Gemini 4 Argon generate images or videos?
Google's announcement describes image and video understanding, not generation. Google's image model is Nano Banana and its video model is Veo.
How much will Gemini 4 Argon cost?
Google lists $2 per million input tokens and $10 per million output tokens at launch, then $4 and $20.
Is Gemini 4 Argon available on LazyKiwi?
No. LazyKiwi is an image and video generator; for Google's image model, use Nano Banana in the image generator.
Is "Gemini Argon" the same as Gemini 4 Argon?
Yes. "Gemini Argon", "Gemini Argon 4" and "Gemini 4 Argon" all refer to the first Gemini 4 model, announced by Google DeepMind on September 30, 2026.
Who can use Gemini 4 Argon today?
Only trusted cyber defenders and testers. Google says paid Gemini API customers and Google AI Ultra subscribers come next, without a date.
How long can Gemini 4 Argon's answers be?
Google lists outputs of up to 1 million tokens, aimed at long coding and document work.
What can Gemini 4 Argon take as input?
Text, images and video. Output is text and code; Google's announcement does not describe image or video generation.
Make images with Google's image model
Argon writes and reasons; for pictures, try Nano Banana in LazyKiwi's image generator.
More models
Image models you can run in LazyKiwi today.