# Gemini 4 Argon opens up: Google’s frontier model is now on sale to everyone who pays

URL: https://ai-pulse.aglaganova.com/blog/gemini-4-argon-opens-up-googles-frontier-model-is-now-on-sale-to-everyone-who-pays
Published: 2026-10-14
Author: Roman Aglaganov, Full-stack developer
Category: Models
Region: Global
Tags: coding, science, benchmarks

> After a month in the hands of vetted cyber defenders, Google’s most capable model is available through its API, the Gemini app and AI Mode in Search. It is aggressively priced, but there is no free tier.

![Illustration: a tall stack of thin diamond-shaped layers pierced from top to bottom by one red line](https://ai-pulse.aglaganova.com/api/media/file/argon-1.png)

Google’s newest frontier model spent its first weeks behind a velvet rope: only a group of “trusted cyber defenders” in its Fairwind Program could use it. On 13 October, Google opened Gemini 4 Argon to developers and businesses worldwide, and to paying consumers.

## Who can use it, and for how much

**Developers and businesses:** through the paid tier of the Gemini API in Google AI Studio and on Google’s Gemini platform.

**Consumers:** Google AI Pro and Ultra subscribers get it in the Gemini app and in AI Mode in Search. Ultra subscribers can also use it in Antigravity, Google’s agent-first coding tool.

**Price:** an introductory $2 per million input tokens and $10 per million output tokens, with cached input 95% cheaper. Gizmodo reports the rates double in December, which would still leave it well below OpenAI’s GPT-6 Astra at $10 and $50.

There is no free version, which Gizmodo argues will limit its mainstream appeal.

## What Google says it can do

Google pitches Argon at long, multi-step “deep work” in coding, finance, law, cybersecurity and science, and says it ranks first on the Vals Index, a benchmark built around economically important tasks. That is Google’s own claim. Its examples are more concrete:

**Code:** Argon helped the Chrome team separate core webpage handling from a legacy folder of about 6 million lines, finishing a refactor across hundreds of files in weeks. It also made Google’s small open Gemma 4 E2B model 2.85 times faster to serve, with identical output.

**Science:** working with Professor Matt Landreman of the University of Maryland, Google says the model helped find a broader class of counterexamples to Grad’s conjecture, a 1967 hypothesis about the magnetic fields in stellarator fusion reactors.

**Security:** Google is expanding the Fairwind Program to hundreds more partners, so defenders keep earlier access to its most advanced cyber capabilities.

## Why it matters

The launch had a bumpy run-up. Gizmodo notes reports of scepticism among some Google employees about Argon’s coding, after a reportedly delayed Gemini 3.5 Pro and the loss of senior researchers. Pricing well below rivals is Google’s answer. Whether developers switch will depend less on the benchmark table than on how Argon copes with their own codebases.

## Sources

- [Google: Gemini 4 Argon, built to solve your hardest problems](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/)
- [Google: Gemini 4 Argon is now available for developers](https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-4-argon/)
- [Gizmodo: Google’s Gemini Argon 4 is now available worldwide](https://gizmodo.com/googles-gemini-argon-4-is-now-available-worldwide-2000825497)