Google Launches Gemini 4 Argon With Advanced AI Reasoning Capabilities

Google has launched Gemini 4 Argon globally, introducing its latest reasoning model for developers, businesses, and subscribers. Google says Argon handles coding, legal, financial, cybersecurity, and scientific workflows.

Ad

Ad

Gemini 4 Argon Reaches Paid Users And Developers

According to an October 13 announcement, Google offers Argon through the Gemini app and Search AI Mode to Pro and Ultra subscribers. Developers can access Argon through Google AI Studio and the Gemini API, while businesses use Google Cloud. Meanwhile, Google AI Ultra customers can run the model through Antigravity, its agent-based software development environment.

The company introduced pricing of $2 per million input tokens and $10 per million output tokens. Google’s developer documentation says both rates double after November 30, when the introductory period ends.

According to its specifications, it has a context window of 1,048,576 tokens and multiple levels of reasoning effort. While Google’s output is text, Google says that Argon is capable of accepting text, images, video, audio, and PDF documents.

Ad

Ad

Google Reports Gains In Coding And Reasoning

In internal tests, Google reported over 35% more agentic code submissions through Antigravity. The firm claimed teams have been able to decrease code rollbacks by almost a third when using Argon for engineering. For example, Google reported a 40% reduction in session preparation time for its Agent Development Kit.

In its own statement, Google said its Argon earned the top spot on the Vals Index, which evaluates the performance of professionals in artificial intelligence. The company said that the areas of finance, legal work, coding, and tax work were among those that performed well. It has a model card that outlines safety testing and explains how the model is tested and its limitations.

Gemini’s Argon 4 launch comes five days after Anthropic released Claude Haiku 5.5. As it was earlier reported by Coingape, Anthropic introduced Claude Haiku 5.5 for high-volume tasks, including customer support. That smaller version of the company was $0.10 per million input tokens for prompts of less than 100,000 tokens. 

Research And Cybersecurity Expand Argon’s Applications

On the cybersecurity front, Google says Argon has enabled researchers to detect and fix software defects in a safe environment. Through Google’s Fairwind Program, Wiz explored areas of vulnerability in healthcare and government. The program will roll out to hundreds more vetted security partners beginning October 13, Google said.

In the field of scientific research, Google announced that Argon helped speed molecular modeling calculations and enabled quantum computing experiments. Researchers at Isomorphic Labs used the model to review patents and scientific documents for drug discovery work.

Google said that the safety assessments included those for biological misuse, cyber threats, and unauthorized use of autonomous agents. The company said that in testing, it reduced the rate of successful prompt-injection attacks to 0.7% of 15 adaptive attempts. Google opened applications for 10,000 scientists to receive six months of free Antigravity access. 

Google’s Gemini 4 Argon launch comes after Anthropic committed $150 million to federal AI research. Anthropic announced the funding on October 8 as part of its participation in the US Genesis Mission.

For more investment opportunities, retail investors can explore pre-IPO tokens offering exposure to major private tech companies.