How to Use Gemini 4 Argon: Gemini App, AI Studio, API and Antigravity (2026 Guide)

Want to know how to use Gemini 4 Argon? Google’s new flagship model became widely available on 13 October 2026, two weeks after it was first announced for a small group of cybersecurity partners. You can now use it in the Gemini app, in Google AI Studio, through the Gemini API and in Google’s Antigravity coding platform, but each route has different plan requirements. This guide explains which one you can use, walks through each method step by step, and shows what Argon costs before its introductory API prices double on 1 December.

Access rules, model details and prices below were checked on 14 October 2026 against Google’s announcements, the Gemini API documentation and Google’s pricing page. We have not run our own benchmark tests on Argon; performance figures are attributed to the company or organisation that published them. API prices are in US dollars and exclude VAT.

What Is Gemini 4 Argon?

Gemini 4 Argon is Google’s most capable model so far. Google DeepMind describes it as its “frontier model for real-world coding, enterprise knowledge work, and cyber defense”, built for long, multi-step workflows such as large code migrations, financial and legal analysis, and finding and patching security vulnerabilities.

Key facts from Google’s developer documentation:

• Model ID: gemini-4-argon (listed as a stable model).
• Context window: 1,048,576 tokens, shared between input and output.
• Output: 65,536 tokens by default for a normal request, or up to about 1 million tokens with Google’s “long decode” option.
• Inputs: text, images, video, audio and PDFs. Output is text only; it does not generate images or audio.
• Thinking levels: low, medium (the default) and high.
• Tools: function calling, code execution, Google Search and Google Maps grounding, URL context, file search, structured outputs and computer use (in preview). The Live API is not supported.

Google first announced Argon on 30 September, but initially released it only to trusted cyber defenders through its Fairwind Program. The wider rollout to subscribers and developers began on 13 October.

Who Can Use Gemini 4 Argon?

Argon is not available on free plans. Here is what each route needs, based on Google’s 13 October announcements:

Google’s API pricing page states that “Gemini 4 Argon is not available on the Free tier or on the Google AI Pro plan”, so a Google AI Pro subscription gives you Argon in the Gemini app but not in AI Studio or Antigravity. In the UK, Google AI Pro costs £18.99 a month and AI Ultra starts at £79.99. Our Gemini pricing guide compares every plan.

Is Gemini 4 Argon available in the UK and Europe?

Google’s announcements do not list country restrictions. Sundar Pichai described the developer release as “available globally today for developers and enterprises”, and the Gemini app announcement simply says Pro and Ultra subscribers can access it. If you are in the UK or the EU and do not see Argon yet, the rollout may not have reached your account, so check again over the next few days before assuming it is blocked.

Method 1: Use Gemini 4 Argon in the Gemini App

This is the easiest route and works on a phone.

1. Check your plan. Sign in with the Google account that has Google AI Pro or Ultra.
2. Open Gemini at gemini.google.com or in the Gemini app on Android or iPhone, and update the app if you are on mobile.
3. Choose the model. Use the model selector in the message box and pick Gemini 4 Argon. Google’s screenshots show Argon selected in the composer alongside new Chat and Task toggles.
4. Give it a substantial task. Argon is built for long, multi-step work, so it is most useful for things like analysing a long report, planning a project or reviewing code.
5. Try Task mode for longer jobs. Google says a toggle from chat to tasks lets Gemini keep working when you are away from your computer, across Google apps and more than 40 connected apps.

Google says Argon in the app answers more directly, with “less hedging, less repetition, and fewer clichés”, and shares progress more often on long tasks. Usage still counts against your plan’s limits, which refresh every five hours up to a weekly cap; AI Pro has four times the standard limit and Ultra 5x or 20x the Pro limit. If you hit a limit, Google lets subscribers keep chatting with a lighter Flash-Lite model. For automating multi-step jobs in the app, our Gemini Agent guide explains how tasks work, and our guide to Gemini Skills shows how to save reusable instructions.

Method 2: Try Gemini 4 Argon in Google AI Studio

Google AI Studio is a browser-based playground for testing prompts before you write code.

1. Go to aistudio.google.com and sign in with a Google AI Ultra account, or with an account linked to a paid Gemini API project.
2. Open a new chat in the Playground and select Gemini 4 Argon from the model list.
3. Set the thinking level in the run settings. Start with medium, and use high for difficult reasoning or coding.
4. Upload files such as PDFs, images or video if your task needs them, then run your prompt.
5. When you are happy, use the code export option to copy the request into your own project.

AI Studio also includes Build, which Google’s Android team says Ultra subscribers and paid API users can use with Argon to create app prototypes.

Method 3: Call Gemini 4 Argon Through the Gemini API

Step 1: Set up a paid API key

Create an API key in Google AI Studio and enable billing on the linked Google Cloud project, because Argon is not on the free tier. Store the key in an environment variable called GEMINI_API_KEY:

export GEMINI_API_KEY="YOUR_API_KEY"

Step 2: Install the Python SDK

pip install -U google-genai

Step 3: Send your first request

Google’s current quickstart uses the Interactions API. This example, adapted from Google’s documentation, sets the thinking level to high:

from google import genai

client = genai.Client()

interaction = client.interactions.create(
    model="gemini-4-argon",
    input="Review this function for bugs and suggest a safer version: ...",
    generation_config={"thinking_level": "high"},
)
print(interaction.output_text)

The same request with curl:

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "gemini-4-argon",
    "input": "Summarise the key risks in this contract: ...",
    "generation_config": {"thinking_level": "medium"}
  }'

Step 4: Choose the right settings

• Thinking level: low for quick answers, medium for most tasks, high for hard reasoning. Thinking tokens are billed as output, so higher levels cost more.
• Thought summaries: set "thinking_summaries": "auto" in the generation config if you want to see a summary of the model’s reasoning.
• Long outputs: a normal request returns up to 65,536 tokens. For very long outputs, such as large code migrations, read Google’s long decode documentation before relying on the 1 million token limit.
• Grounding: enable Google Search grounding for up-to-date answers. Google includes 5,000 grounded requests a month across Gemini 3 and newer models, then charges $14 per 1,000.
• Batch and Flex: for jobs that do not need an instant reply, Batch and Flex processing cost half the standard rate.

Argon is also available to businesses through Google Cloud’s Gemini Enterprise Agent Platform and Gemini Enterprise, under the same model ID.

Method 4: Use Gemini 4 Argon in Antigravity for Coding

Google Antigravity is Google’s agent-first development environment. Argon is available there only to Google AI Ultra subscribers.

1. Download the latest Antigravity desktop app, or install the Antigravity terminal CLI.
2. Sign in with your Google AI Ultra account.
3. Open your project folder and select Gemini 4 Argon as the model.
4. Describe the task, such as “migrate this module to TypeScript and update the tests”, and review the agent’s plan and changes before accepting them.

Android developers can add Google’s Android tools as a plugin. Google’s Android team lists three ways: enable android-cli-plugin in Antigravity Desktop under Settings > Customizations, run agy plugin install android-cli-plugin in the Antigravity terminal, or sign in to Antigravity inside the latest Android Studio Canary. Google also says Antigravity’s coding interface is coming to the Gemini app for macOS.

Any coding agent can run commands and change files, so work on a branch and check permissions first. Our AI agent security guide covers the main precautions.

Gemini 4 Argon Pricing

Argon has a single rate per token type, with no higher price for long prompts. Introductory prices apply until 30 November 2026; from 1 December they double.

Caching is implicit only, meaning Google applies the discount automatically when it reuses part of your prompt; explicit caching is not supported.

Worked example: a request with 100,000 input tokens and 10,000 output tokens costs about $0.30 at the introductory standard rate ($0.20 input plus $0.10 output), or about $0.60 from 1 December. Our own calculation, before VAT.

Independent testers note that Argon can be wordy. Artificial Analysis reports that it generated 110 million tokens on its Intelligence Index, compared with a median of 85 million, so set clear length limits in your prompts to control output costs.

What Is Gemini 4 Argon Good At?

Google’s own launch figures put Argon at 77.9% on DeepSWE v1.1, first on Zapier’s AutomationBench at 51.3% and 91.7% on the LVBench long-video test. Independent results are more mixed. On Vals AI’s index, Argon ranked first at 68.90%, but Artificial Analysis scores it 53 on its Intelligence Index, level with OpenAI’s GPT-6 Astra and behind Anthropic’s Claude Opus 5.5 at 58, according to VentureBeat. Our Gemini 4 Argon vs Claude Opus 5.5 comparison puts these benchmarks side by side.

Based on the published results, Argon looks strongest for:

• Financial and legal analysis: it ranks first on Vals AI’s Finance Agent benchmark (65.40%).
• Code migration and review: Google says its engineers use it daily for debugging and migrations, including moving C and C++ code to Rust.
• Long video and documents: its 1 million token window and video input suit long recordings, reports and codebases.
• Workflow automation: its AutomationBench result suggests strength in multi-app tasks.

Tips for Better Results

1. Match the thinking level to the task. High is slower and more expensive; it is not needed for simple questions.
2. Give it everything at once. Upload the full report, codebase or transcript rather than pasting fragments.
3. Ask for a plan first on large coding or analysis tasks, then approve it before execution.
4. Set output limits such as “answer in under 300 words” to reduce verbosity and cost.
5. Check every important claim. Even frontier models make mistakes, especially with figures and citations.
6. Move off introductory pricing deliberately. If you build an app now, budget for the 1 December price rise.

Limitations to Know

• No free access: free Gemini users and the free API tier cannot use it.
• Coding tools need Ultra: AI Pro subscribers cannot use Argon in Antigravity or AI Studio.
• Prices double on 1 December 2026.
• No image or audio output and no Live API support.
• Usage limits apply in the Gemini app, even on Ultra.
• Most performance claims come from Google. Independent rankings vary by benchmark.

If Gemini does not load or shows errors, our Gemini troubleshooting guide covers common fixes.

Frequently Asked Questions

Is Gemini 4 Argon free?

No. It requires Google AI Pro or Ultra in the Gemini app, Google AI Ultra for Antigravity, or the paid tier of the Gemini API.

What is the Gemini 4 Argon model ID?

The API model ID is gemini-4-argon.

How much does the Gemini 4 Argon API cost?

$2 per million input tokens and $10 per million output tokens until 30 November 2026, then $4 and $20 from 1 December. Batch and Flex cost half as much.

Can I use Gemini 4 Argon with Google AI Pro?

Yes, in the Gemini app and AI Mode in Search. Antigravity and AI Studio access require Google AI Ultra or a paid API account.

What is the Gemini 4 Argon context window?

1,048,576 tokens, shared between input and output, with outputs of up to about 1 million tokens using long decode.

When was Gemini 4 Argon released?

Google announced it on 30 September 2026 for selected cyber defence partners and made it widely available to subscribers and developers on 13 October 2026.

Conclusion

The quickest way to use Gemini 4 Argon is in the Gemini app with a Google AI Pro or Ultra plan. Developers should use a paid Gemini API key with the gemini-4-argon model ID, and Ultra subscribers get the most complete coding experience in Antigravity. If you plan to build on the API, test now while introductory prices last, set clear output limits, and budget for the price doubling on 1 December.

Sources: Google – Gemini 4 Argon announcement; Google – Gemini 4 Argon for developers; Google – Gemini 4 Argon in the Gemini app; Gemini API – Gemini 4 Argon model page; Gemini API pricing; Gemini API – thinking; Gemini API quickstart; Android Developers Blog – Argon in Antigravity and Android Studio; Google Cloud – Gemini 4 Argon; Gemini Help – usage limits; VentureBeat; OfficeChai; Artificial Analysis; Vals AI.