reasoningunverifiedHand-written lesson

Google

Gemini 4 Argon

google/gemini-4-argon

Google's flagship reasoning model for complex software engineering, deep math and science, and autonomous multi-step agents. Takes text, images, files, audio, and video in one request.

RoleThe brain when inputs are not just text

$2

In / M

$10

Out / M

1.0M

Context

AA intelligence index
50.4
Inputs
audio, file, image, text, video
Outputs
text
Tools
Listed
Structured output
Listed
Reasoning
Listed
Weights
Not verified
Listed
Oct 13, 2026
Live from the OpenRouter listing ↗

01 · What it's good at

Use it for this.

Three jobs this model is built for, and the ones to give to something else.

01

Reasoning over mixed media: video, audio, screenshots, and PDFs together

02

Complex software engineering and multi-step agents

03

Math and science problems that need a long chain of reasoning

Skip it for

  • Cheap bulk work. Gemini 3.8 Flash covers most of the same inputs for less
  • Word-level timestamps. Use a transcription model

02 · Example tasks

What you'd actually send it.

Real inputs, not a benchmark. Paste one into the snippet below to try it.

You send

Here is a 40-second screen recording of the checkout bug and the component source. Describe what the user did, where it breaks, and the likely fix.

Why this model

Video input means no one has to write the repro steps by hand.

03 · Your stack

Where it sits: Plan & reason.

Make it the reasoning step when your pipeline receives recordings, screenshots, or long documents. Send media straight in instead of chaining OCR or transcription first, and keep a dedicated transcription model only where you need timestamps.

04 · Wire it up

One request. Paste and run.

Uses OpenRouter, so one key works for every model on this site. Set OPENROUTER_API_KEY and swap the prompt for one of the tasks above.

// OpenRouter speaks the OpenAI Chat Completions format.
const res = await fetch("https://openrouter.ai/api/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "google/gemini-4-argon",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "This screenshot shows our checkout error. What is the user seeing, and what is the most likely cause in a React app?"
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "https://example.com/checkout-error.png"
            }
          }
        ]
      }
    ],
    "reasoning": {
      "effort": "high"
    }
  }),
});
const data = await res.json();
console.log(data.choices[0].message.content);

05 · Watch out for

Gotchas.

  • Effort levels on this listing are low, medium, and high only.

  • Audio and video inputs are large. Watch usage in the response before you run it on a backlog.

Sources

Lessons are written from each model's listing and the linked docs. We don't publish our own benchmark numbers; the index above is Artificial Analysis, as shown on OpenRouter.

Keep learning

Same board, other models.