Google makes Gemini 4 Argon available to developers, businesses and paid subscribers
Google has begun rolling out its most advanced AI model, Gemini 4 Argon, more widely, making it available to developers, businesses and paid subscribers, while users of the free version still cannot access it.
اضافة اعلان
Google unveiled Gemini 4 Argon in late September. Technically, it appeared to be the company’s strongest response yet to compete with advanced models developed by OpenAI and Anthropic.
However, the launch faced a major obstacle: the vast majority of users were unable to try the model themselves.
When it was first announced, access to Argon was limited to a small group of testers, trusted partners and government entities, while Google continued conducting the necessary tests before a broader release.
Two weeks after its initial launch, the company finally decided to expand access and make the model available to new categories of users worldwide.
Gemini 4 Argon is now available globally to developers and businesses through the paid tier of the Gemini API, AI Studio and Google’s Gemini platform.
Subscribers to the Google AI Ultra plan can also use the model through the Antigravity platform. Meanwhile, subscribers to the Pro and Ultra plans can access it through the Gemini app and Google Search’s AI Mode.
Google expands access to its most powerful model
Argon remains primarily designed for complex tasks requiring advanced analysis and reasoning, rather than simply providing quick answers to basic questions.
Google promotes the model as suitable for a wide range of specialized fields, including programming, finance, legal work, cybersecurity and science, as well as tasks requiring extensive processing and in-depth analysis before reaching conclusions.
The company says Argon currently ranks first on the Vals benchmark, which measures AI model performance across several fields, including finance, programming, legal work and taxation.
Google’s move goes beyond expanding access. The company has also introduced introductory pricing for using the model through its APIs in an effort to attract developers and businesses seeking to integrate its capabilities into their services and systems.
The introductory price is $2 per million input tokens and $10 per million output tokens.
Google has also announced discounts of up to 95% on the cost of cached input tokens. These are data that can be reused without being fully processed each time, potentially reducing operating costs for applications that rely on the model repeatedly.
Notably, Google had previously outlined which groups would gain access to Argon first. From the outset, it said customers using paid API services and subscribers to the Google AI Ultra plan would be among the first to benefit from the model.
Better coding performance and fewer errors
Google now has more practical examples of Gemini 4 Argon in use than it did when the model was announced in September, and the company is using these examples to highlight its capabilities in real-world working environments.
According to Google, the model helped increase code submissions produced by the company’s AI agents by more than 35%, while reducing code rollbacks by about one-third.
Rollbacks refer to situations in which code must be restored to an earlier version after problems or errors are discovered in new changes.
Chrome engineers also used the model in a large-scale code refactoring effort involving hundreds of files in the Chromium project, whose codebase contains approximately 40 million lines of code.
These are considered complex technical tasks because they require modifying large portions of code while preserving their connections and ensuring that different components continue to function correctly.
Google presents this experience as an example of Argon’s ability to help maintain large software projects involving extensive networks of interconnected files and code instructions.
Applications in cybersecurity and scientific research
Gemini 4 Argon’s applications have extended beyond software development. Google has also used it to improve its suite of tools for developing AI agents, which help developers build AI systems capable of carrying out tasks more independently.
The model was also used to accelerate a software package specializing in quantum chemistry known as XTB, a tool used in scientific calculations and simulations involving molecular properties.
Other applications included improving reasoning processes in the Gemma 4 model, helping advance the mechanisms used to run AI models and generate their responses.
In addition, researchers have drawn on Argon’s capabilities for work involving cybersecurity and scientific research, reflecting Google’s efforts to apply the model in areas beyond conventional conversational applications.
These applications form part of the company’s efforts to demonstrate that its advanced models can address technical and scientific challenges requiring high levels of accuracy and analysis.
Free users remain excluded
Despite the significant expansion of Gemini 4 Argon’s availability, access remains limited for ordinary users who rely on the free version of Gemini.
While developers and businesses can use the model through paid services, individuals can access it by subscribing to either the Google AI Pro or Ultra plan.
Argon therefore remains an exclusive feature for paid subscribers for now, despite moving from limited testing to a broader global rollout.
The expansion marks another step in Google’s strategy to strengthen its position in the advanced AI market by combining powerful models with practical applications in programming, scientific research and business.
However, free Gemini users will have to wait for any future decision by the company to expand access, as the current announcement does not include this group.
Source: Al-Ghad.
