Gemini 4 Argon Released: Google's Most Powerful AI for Solving Complex Problems [Watch]

The Gemini 4 Argon AI model, featuring deep reasoning, is being applied to complex projects in engineering, programming, and finance.

Argon's Gemina 4 artificial intelligence model is applied to complex engineering, programming, and finance projects with deep reasoning.

Become a special member of Arjan Ambassador in order to benefit from and access special facilities and different departments in Arjan Ambassador.

Google has officially launched its latest and most advanced artificial intelligence model named Gemina 4 Argon. This model is designed to perform deep work on complex and real-world tasks in areas such as programming, finance, legal, cyber security, and basic science. The processing depth and high accuracy of Argon in multi-stage tasks have made this model surpass other competitors in specialized benchmarks and receive the first rank in the Vals index; An index that measures the performance of models in sensitive economic sectors such as finance, programming, rights and taxes.

The new Gemina model is now globally available to developers and companies through API in Google Artificial Intelligence Studio and Gemina platform. Also, subscribers of the Google AI Ultra service have access to it on the Google Antigravity platform, and Pro and Ultra subscribers have access to it in the Gemini application and the artificial intelligence mode of the search section. The release price of this model for developers is set at $2 per 1 million input tokens and $10 per 1 million output tokens, and a 95% discount is included for cached input tokens.

This initial price is approximately one fifth of the cost of competitors such as GPT-6 Astra and Fable 5 family models; Of course, this tariff is an initial discount and will double at the end of November, although even after doubling it will still be cheaper per token than competing models.

Google claims that this model has been able to defeat fierce market competitors, including OpenAI's GPT-6 Astra, as well as the Claude Fable 5.1 and Claude Opus 5.5 models from Anthropic in a series of specialized benchmarks for programming, cyber security, science, finance and law.

Google's internal teams have been using the Argon model for the past few weeks to solve their toughest engineering challenges. The engineers of this company have been able to develop stable and high-quality codes at a much higher speed; So that the use of argon in the Antigravity environment has increased automatic code registration by more than 35% and reduced the number of times to restore codes to the previous version by nearly a third.

In one major project, Argon helped the Chrome browser team to separate the web page processing core from an old 6 million-line folder within Chromium's 40 million-line code base. By applying multiple simultaneous rewrites, the model completed architectural reconstruction in hundreds of files, completing a task that would have taken months in a matter of weeks. This change allows Chrome's billions of users to switch tabs live between sidebars and floating windows without reloading.

In addition, Google has made other extensive uses of argon in Alphabet products; including the optimization of the YouTube comments section, which has reduced the processing power required to run it by half. Also, Wing, Alphabet's drone package delivery arm, has used this artificial intelligence to troubleshoot and debug the flight physics simulator of its drones.

In addition to Chrome, Argon was able to optimize the Google Agent Development Kit Python code base in a few hours, reducing the preparation time of long sessions in memory by 40%. Researchers in the field of quantum chemistry also made the famous software package xTB (three-dimensional modeling package of molecules) 2.3 times faster in predicting structures by using argon. Also, in Gemma's open source models, Argon helped achieve 2.85x processing acceleration on GPUs without quality loss. In visual and 3D processing, this model has the ability to create realistic 3D objects, interactive animations and even video games directly on the web.

Gemini 4 Argon model, in addition to being skilled in analytical and programming tasks, also has a fun potential in creating interactive projects; So that a user in X asked it to design a magical chess game (similar to Harry Potter's wizarding chess) and Argon's high ability to understand the laws of physics and 3D modeling reminded him of the nostalgic Battlechess game on old Amiga computers. The following experience was created by Gemina 4 for this user:

The release of Gemini 4 Argon is considered the first major unveiling of Google in the category of frontier models after the introduction of Gemini 3 in late 2025. In recent months, in the wake of Entropic and OpenAI's impressive advances, there have been doubts about Google's ability to maintain its leadership position. However, the release of Argon, which took place only one day after the unveiling of the Muse Wonder Full model by Meta, showed that Google has no intention of backing down from the heavy AI competition.

My professional career in the world of technology goes back almost 10 years; The path that started with hardware, continued with software, and now continues as a technology reporter in Sefir Arjan. I am Javad and more than three decades have passed and in addition to the world of technology, I love movies and cinema, classic rock music and Real Madrid.