Google has officially begun the massive rollout of its new flagship artificial intelligence model, named Gemini 4 Argon. The presentation of this advanced development has drawn immense attention from the IT community, as the corporation positions the system as a qualitative leap in generative intelligence, reasoning, and code writing. The launch of the technology comes against the backdrop of fierce competition in the global AI market, where each new iteration must prove its superiority not just in words, but in practice.
Technological Features and Capabilities of Argon
The characteristics of the top-tier AI remain unchanged, yet developers have focused heavily on deepened logic and safety. Google references recent benchmarks as proof of advanced reasoning and code agent capabilities. Additionally, the model is trained to actively reject malicious instructions, and a special built-in monitor can instantly halt processing if it detects an unpredictable or potentially dangerous workflow. Such measures are designed to minimize the risks of autonomous task execution.
The tech giant has already cited several striking examples of successful internal use of Argon within the corporation. Specifically, while optimizing YouTube comments, the neural network discovered redundant emoji rendering, the elimination of which reduced server load by an impressive 45 percent. In the Wing project drone physics simulator, the artificial intelligence found critical bugs that had remained in the code since 2018. The model also demonstrated excellent results in robotics, showing a noticeable step forward in spatial reasoning for three-dimensional environments.
Contradictory Data
Despite impressive official demonstrations and internal corporate tests, serious controversies have flared up around the release of Gemini 4 Argon. Some specialized publications and independent experts criticized the novelty in the very first days after launch, claiming, with reference to their own tests, that the model allegedly demonstrates inflated benchmark results due to task-specific optimization and sometimes stumbles when writing complex production code. Google representatives are currently refraining from detailed comments regarding the criticism, insisting on the objectivity of standard metrics.
Access Conditions and Pricing Policy
The new model is distributed on a commercial basis, confirming the corporation's concept of the high value of advanced computing. In the familiar Gemini consumer app, the model becomes available under the AI Pro subscription for $20 per month and AI Ultra, starting from $99 per month. In specialized developer and coding products such as Antigravity and AI Studio, Gemini 4 Argon is integrated exclusively on the most expensive AI Ultra tier due to massive computational costs.
For less complex and routine tasks, the company wisely retains the previous version, Gemini 3.1 Pro. For developers working via the paid API, the temporary token cost is set at $2 per million input tokens and $10 per million output tokens. Furthermore, a special 95 percent discount on cached tokens applies until November 30. After the promotion ends, standard rates will double—to $4 for incoming and $20 for outgoing tokens.
