Google debuts Gemini 4 Argon AI with intro pricing, topping Vals Index in finance, coding, and legal tasks

Google has made Gemini 4 Argon available globally for developers and enterprises through its API in AI Studio and the Gemini platform. The model is also accessible to Google AI Ultra subscribers in Google Antigravity, and to Ultra and Pro subscribers in the Gemini app and AI Mode in Search, according to the company’s announcement.

Model capabilities

The release introduces a frontier model designed for complex, real-world tasks. Google states that Argon excels in domains including coding, finance, legal, cybersecurity, and science. Its design focuses on depth and precision across multi-step tasks, allowing it to outperform other frontier models on benchmarks across these sectors.

Performance benchmarks

Argon ranks number one on the Vals Index, which measures model performance in critical economic sectors such as finance, coding, legal, and tax work. This benchmarking highlights the model’s capability to handle intricate workflows that require high accuracy and sustained logical reasoning over multiple steps.

Access tiers

The model is currently available in the following access tiers:

• Developers and enterprises via the API in AI Studio and the Gemini platform.
• Google AI Ultra subscribers in Google Antigravity.
• Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search.

Usage examples

• Google engineers used Argon to refactor Chrome’s legacy codebase, completing months of work in weeks.
• The model optimised the Google Agent Development Kit, cutting in-memory session prep time by 40%.
• Researchers used Argon to speed up the xTB quantum chemistry package by up to 2.3x.

What it means for organisations

Teams usually need to assess whether integrating this model aligns with existing security and compliance frameworks. Engineering leaders should evaluate the risks associated with automated code refactoring and ensure that automated changes undergo rigorous review. Procurement teams must verify that data handling practices meet regulatory requirements, while risk managers should consider the implications of relying on external AI for critical infrastructure optimisation.

Background: APIs

APIs allow your application to request data from a server on demand. Webhooks allow a server to send data to your application when specific events occur. APIs pull data; webhooks push it. Choose APIs for user-initiated actions and webhooks for asynchronous event notifications.

Read the full guide: APIs vs Webhooks: Choosing the Right Integration Pattern

What to do and how to stay safe: Google

• Establish clear guidelines for using AI-generated code in production systems.
• Monitor API usage and costs to prevent unexpected expenditure during the introductory pricing phase.
• Verify that any automated security patches or code refactoring undergo manual review before implementation.
• Ensure that data shared with the model complies with organisational privacy and security policies.

Step-by-step guide: Prevent Prompt Injection Attacks: Practical Defences for AI Systems

General security guidance from the TechSec Observer newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What is the introductory pricing for Gemini 4 Argon?

The model launches at $2 per million input tokens and $10 per million output tokens, with cached input tokens at 95% off.

Which sectors does Gemini 4 Argon excel in?

It excels in coding, finance, legal, cybersecurity, and science, ranking first on the Vals Index for these domains.

How is Gemini 4 Argon being used at Google?

Google engineers are using it to refactor large codebases, optimise Python frameworks, and speed up quantum chemistry packages.