Gemini 4 Argon: What Google's New Model Means for Creators

Google announced Gemini 4 Argon on September 30, 2026. It thinks, codes and watches video, but it doesn't make images or video. Here's what creators can use today.

3 min read

Gemini 4 Argon: What Google's New Model Means for Creators

Illustration made on Lumeta with GPT Image 2.5Make one like it

Google DeepMind announced Gemini 4 Argon on September 30, 2026, and it's already everywhere in your feed. It's the first model of the Gemini 4 generation, built for long coding jobs, business knowledge work and cyber defense. If you make images, videos or music, here's the short version: Argon is a thinking model, not a making model. The tools that actually create your visuals are already here.

What Gemini 4 Argon actually does

  • It writes a lot at once. With long decoding, up to 1 million tokens in a single answer. That's a whole script bible, not a paragraph.
  • It understands video. It takes text, images, video, audio and PDFs in. Google reports 91.7% on LVBench, a test of understanding long videos.
  • It leads most of Google's benchmarks. First on 13 of the 19 tests in Google's own table, against GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1.
  • It's available now. Since October 13, 2026, Google AI Pro and Ultra subscribers can use it in the Gemini app, and developers can call it as gemini-4-argon on the paid Gemini API.

Where Argon fits in a creator's workflow

Think of Argon as the sharpest person in your writers' room. It's the model to hand a long brief: plan a 20-episode series, write every shot prompt for a 60-second ad in one pass, or watch your rough cut and list every continuity slip with timestamps.

The pictures and clips still come from generation models. Google's own are Veo 3.1 for video with sound, Gemini Omni for short clips you edit by chatting, and Nano Banana 2 for images. All three are on Lumeta right now, next to Seedance 2.5, Kling O3 and MiniMax H3.

Veo 3.1Google's video model, with sound built in Gemini OmniMake a short clip, then change it by chatting Nano Banana 2Google's image maker and editor

A workflow you can use today

Argon is a great fit for step one, and any strong language model can handle it too.

  1. Plan

    Ask a language model for a one-line concept, a shot list and one prompt per shot.

  2. Lock the look

    Make one clean still with Nano Banana 2 and use it as the start frame for every shot.

  3. Generate

    Run each shot in Veo 3.1 or Gemini Omni, then try the same prompt in Seedance or Kling. Keep the best take.

  4. Finish

    Sharpen it with Upscale Video, add a soundtrack with Music Generator and match a voice with Lip Sync.

Want the full details?

We also put together two independent guides to Argon. For developers, gemini4argon.com covers the API, Google's full benchmark table and pricing. For everyone else, googleargon.com explains Argon in plain English, including whether it's safe. Google has since announced a stronger model, Gemini 4 Carbon, and our Gemini 4 Carbon guide tracks when you can use it. None of these sites is affiliated with Google.

Can Gemini 4 Argon make videos?

No. Argon takes text, images, video, audio and PDFs in and writes text out. For video, use Google's Veo 3.1 or Gemini Omni, both available on Lumeta.

When can I use Gemini 4 Argon?

Now. Since October 13, 2026, Google AI Pro and Ultra subscribers can use it in the Gemini app, and developers can call gemini-4-argon on the paid Gemini API.

Is Gemini 4 Argon on Lumeta?

No. Argon is a text model, available through Google's own Gemini app and API. Lumeta is for making images, video, voice and music, with Google's Veo 3.1, Gemini Omni and Nano Banana 2 among its tools.

Every top AI model, in one place

Nano Banana 2, GPT Image 2.5, Seedance, Veo and more, all on one credit balance. Start free with 35 credits a month. No card needed.

Browse all tools

More from Lumeta

All News