Gemini 4 Argon

Gemini 4 Argon is Google DeepMind's frontier model, announced September 30 and expanded on October 13, 2026 to paid Gemini API developers and Google AI Pro/Ultra app users.

Language ModelAgent Workflowsgoogle

Hosted google API; access remains account and surface specific

Intro through November 30, 2026: $2/1M input and $10 output; after November 30: $4/$20; cache charges follow provider contract

Scoped Terms

🚀Function Overview

Gemini 4 Argon is Google DeepMind's frontier model, announced September 30 and expanded on October 13, 2026 to paid Gemini API developers and Google AI Pro/Ultra app users.

Key Features

Paid Gemini API/AI Studio rollout October 13; app Pro/Ultra and Antigravity Ultra are separate
Text/image/video/audio/PDF input; text output; unified 1,048,576-token context
65,536 default synchronous maximum output; separate long decode up to 1,048,576
Thinking low/medium/high, medium default; Fairwind cyber variant needs separate approval

Use Cases

•Evaluate long-running repository tasks when access is granted.
•Compare document-heavy knowledge work with source verification.
•Assess defensive security workflows within authorized environments.

⚙️Input Parameters

array

Gemini generateContent contents with text or supported media parts

generationConfig

object

Documented output and thinking configuration; ordinary synchronous output is distinct from long decode

💡Usage Examples

Example 1

Input Parameters

{
  "contents": [
    {
      "role": "user",
      "parts": [
        {
          "text": "Summarize the supplied project notes and list uncertain claims."
        }
      ]
    }
  ]
}

Output Results

Illustrative generateContent request; not executed in this review. Read candidates[].content.parts and inspect promptFeedback and usageMetadata before validating the result.

Quick Actions

Technical Specifications

Hardware Type: Hosted google API; access remains account and surface specific
Commercial Use: Hosted use follows provider terms; no open-weight redistribution license is established.
Pricing: Intro through November 30, 2026: $2/1M input and $10 output; after November 30: $4/$20; cache charges follow provider contract