Models
What Is Gemini 4 Argon? Google’s New Frontier Model, Explained
Gemini 4 Argon is Google’s new frontier AI model, out for everyone since October 13, 2026. What it does, the price, who can use it, and whether it is free.
Gemini 4 Argon is the newest and most capable model from Google, and its name is now showing up everywhere: in the Gemini app, in coding tools, and in benchmark charts next to OpenAI and Anthropic. If you are wondering what it actually is, whether you can use it, and what it costs, this page answers that in plain words.
The short story: Google announced Argon on September 30, 2026, first gave it only to a small group of security experts, and then opened it up on October 13. Here is what changed and what it means for you.
What is Gemini 4 Argon?
Gemini 4 Argon (people also type it as "Gemini Argon" or "Gemini 4") is a large AI model made by Google DeepMind. It is the engine behind the Gemini assistant, the way a car engine sits behind the steering wheel you actually touch.
Google describes it as a model built to keep reasoning through "complex, long-horizon workflows." In everyday words: tasks that take many steps and a long time, such as moving a large codebase from one programming language to another, reading a stack of documents and acting on them, or checking software for security holes and fixing them.
The launch post came from Koray Kavukcuoglu, Google DeepMind’s chief AI architect. Google gave a few real examples of Argon at work inside the company. It helped convert old C and C++ code to the safer Rust language, including a Rust video decoder that runs 2.7 times faster than the previous Rust version. Google also says its agents found memory savings expected to free more than 300 TiB of memory once they are rolled out.
Gemini 4 Argon release date
Argon arrived in two steps, which is why some pages still say you cannot use it.
| Date | What happened |
|---|---|
| September 30, 2026 | Google announces Gemini 4 Argon. Access is limited to trusted cyber defenders in its Fairwind Program while Google works with the US government’s voluntary pre-release review. |
| October 13, 2026 | Argon opens up: Google AI Pro and Ultra subscribers get it in the Gemini app, and developers get it through the paid Gemini API (model ID gemini-4-argon) and Google AI Studio. |
| November 30, 2026 | Last day of the introductory API price. |
| December 1, 2026 | API price doubles to the standard rate. |
The security version without cyber guardrails stays separate. Google says that one goes only to vetted defenders through Fairwind.
Is Gemini 4 Argon free?
No. Google’s own API pricing page says plainly that Gemini 4 Argon is not available on the free tier. Inside the Gemini app, Google names only Google AI Pro and Ultra subscribers as having access. Some special tools, like Google AI Studio and the Antigravity coding app, list Argon for Ultra subscribers.
For developers, here is the API price per million tokens. A token is a small piece of a word, so one million tokens is roughly 750,000 words.
| Price per 1M tokens | Until Nov 30, 2026 | From Dec 1, 2026 |
|---|---|---|
| Input | $2 | $4 |
| Output (includes thinking) | $10 | $20 |
| Batch input | $1 | $2 |
| Batch output | $5 | $10 |
For comparison, the standard $4 and $20 rate is the same price Meta charges for Muse Wonder, which also matches Claude Opus 5.5. So the intro price is Google’s way of saying "try us first."
Gemini 4 Argon benchmarks
These are scores Google itself published. Treat them like any launch-day claim: useful, but not yet checked by everyone.
| Test | What it measures | Argon score (Google) |
|---|---|---|
| DeepSWE v1.1 | Long, real-world software engineering | 77.9%, which Google calls a new best |
| LVBench | Understanding long videos | 91.7% |
| AutomationBench (Zapier) | Running business automations | 51.3%, ranked first |
| CWE-bench v1 | Fixing software security flaws | 68%, tied for first |
TechCrunch reported that Google claims Argon scored "significantly higher" than OpenAI’s GPT-6 Astra and Anthropic’s Fable and Opus models across several benchmarks. Independent results are still coming in, and early outside reports are mixed by task. Google also calls Argon its most prompt-injection-resistant model yet. Prompt injection is when hidden text on a web page tries to trick an AI agent into doing something you did not ask for.
Gemini 4 Argon context window and output limit
This is the number that makes Argon different. Google’s developer docs list a shared 1,048,576 token window for input and output combined. By default, one reply can be up to 65,536 tokens. With a "long decode" mode, a single reply can stretch to about 1 million tokens. Earlier Gemini models stopped at 64,000.
Why should a normal person care? A longer output means the model can write a full report, a whole app, or a long translation in one go instead of stopping halfway. Argon also reads text, images, video, audio, and PDFs, though it only writes text.
How to get Gemini 4 Argon
- Gemini app: sign in with a Google AI Pro or Ultra plan and pick Argon. Google says the rollout is gradual, so it may take a little time to show up.
- Developers: use the model ID
gemini-4-argonon a paid Gemini API plan, or test it in Google AI Studio. - Security teams: the version without cyber guardrails is only through Google’s Fairwind Program.
What this means for you
Argon is a real step for Google, especially for long jobs. But look at the past two months: GPT-6 Astra, Claude Opus 5.5, Muse Wonder, DeepSeek V4.1 Flash, and now Argon. Each one comes with its own plan, its own price, and its own list of who gets access first.
For most people, that is a lot of homework just to get an email written or a trip planned. This is where agentic AI helps. Instead of picking a model and a subscription, you hand a goal to expert agents that plan the steps and finish the work.
That is how our app works. Agentic AI gives you a team of ready-made expert agents on your iPhone and iPad, running on frontier models. You can put up to 4 agents on one task, there is no account to create, and your first 3 tasks are free.
Quick answers
What is Gemini 4 Argon? Google’s newest frontier AI model, announced September 30, 2026 and widely available since October 13. It is built for long coding, research, and security tasks.
Is Gemini 4 Argon free? No. Google says it is not on the free API tier, and in the Gemini app it is for Google AI Pro and Ultra subscribers.
When will Gemini 4 Argon be released to the public? It already was, on October 13, 2026, for paying subscribers and paid API users. The guardrail-free security version stays limited to the Fairwind Program.
What is the Gemini 4 Argon price? $2 per million input tokens and $10 per million output tokens through November 30, 2026, then $4 and $20.
Gemini 4 Argon vs Opus 5.5: which is better? Google says Argon leads on several of its chosen tests, including 77.9% on DeepSWE. Independent results vary by task, so the best pick depends on the job.
What is the Gemini 4 Argon context window? About 1 million tokens shared between input and output, with replies of up to about 1 million tokens in long decode mode.
New to the words in this post? Our agentic AI glossary explains tokens, context windows, and frontier models in plain English.
Sources: Google: Gemini 4 Argon · Google: Argon in the Gemini app · Gemini API model docs · Gemini API pricing · TechCrunch · HPCwire
