Google (GOOGL.US) Releases Gemini 4 Argon's 19 Tests, 13 Leading Competitors, Pakistan Maintains $420 Target

The Zhitong Finance App learned that Google's parent company Alphabet (GOOGL.US, GOOG.US) officially announced the next-generation cutting-edge artificial intelligence model Gemini 4 Argon on Wednesday, showing strong competitiveness in various fields such as complex workflows, knowledge-based tasks, cybersecurity defense, multi-modal understanding, and software engineering. The model surpassed competing products of OpenAI and Anthropic in 13 of the 19 benchmarks, drawing attention on Wall Street. BNP Paribas believes that the launch of the new model once again proves that Google is committed to maintaining competitiveness in the cutting-edge AI field and maintaining Alphabet's “outperforming the market” rating and target price of $420.

However, Gemini 4 Argon is currently not fully open to the public. Google said that due to security considerations, at this stage, the model mainly provides services to the US government and some trustworthy cybersecurity agencies, and will gradually expand the scope of openness after improving security protection measures in the future.

Gemini 4 Argon performs brilliantly, leading 13 out of 19 tests

As the latest cutting-edge AI model launched by Google, Gemini 4 Argon achieved leading performance in a number of key competency tests.

According to data released by Google, Argon surpassed OpenAI's Astra and Anthropic's Fable 5.1 and Opus 5.5 in 13 of the 19 benchmarks, showing competitiveness in complex task processing and professional applications. Among them, Argon excels in the fields of complex workflows, knowledge-based work, cybersecurity defense, multi-modal understanding, and software engineering.

Google also said that Argon has been widely used within the company and is mainly used in scenarios such as professional programming tasks, in-depth research, and writing quality optimization.

In terms of commercial pricing, the future usage price of Argon announced by Google is $2 per million input tokens and $10 per million output tokens. For cached input tokens, Google will offer a discount equivalent to 95% of the standard input price.

This means that the actual price of a cached input token will drop to $0.10 per million, helping to reduce the inference costs of reusing contextual information. By providing different price and performance options, Google is expected to further expand the scope of application of the Gemini model among developers and enterprise customers.

Security review delayed full release Google emphasizes rapid iteration

Despite Gemini 4 Argon's impressive performance in performance tests, Google has adopted a relatively cautious launch strategy. Google CEO Sundar Pichai said on social platforms that Argon is already equipped with cutting-edge AI security protection measures and is currently being opened to the US government and some trusted cybersecurity defense agencies participating through the Fairwind Program.

Pichay said that Google will expand the scope of model openness as quickly and safely as possible, and emphasized that it will continue to rapidly iterate and launch more new developments in the future.

Google DeepMind and chief AI architect Koray Kavukcuoglu also said that the company will continue to collect feedback from early test users, further improve the model's security protection mechanism, and strive to open Argon to developers, enterprises and consumers as soon as possible.

Notably, this model was originally scheduled to be released around June of this year, when the market generally expected its name to be Gemini 3.5 Pro. However, due to delays in development progress, Google finally adjusted the product route and switched the R&D focus to Gemini 4 Argon.

According to reports, this extension is mainly related to stricter security requirements and Google's desire to further improve model programming capabilities before official release. This delay has also raised investors' concerns about Google's competitive position with AI. Since the model was postponed, Alphabet's market value has shrunk by nearly 20%, reflecting the market's doubts about whether it can continue to compete with competitors such as OpenAI and Anthropic.

BNP Paribas is optimistic about Google's AI advantage and maintains a target price of $420

The appearance of Gemini 4 Argon received positive reviews from Wall Street analysts.

Nick Jones, an analyst at BNP Paribas, said in a report to customers that it is encouraging that Google continues to advance cutting-edge AI model development. In particular, Pichay made it clear that the company will speed up model iteration, further strengthening the market's confidence in its ability to execute AI strategies.

Jones believes that Google has many advantages in cutting-edge AI competition, including leading AI research capabilities, its own computing infrastructure, extensive product distribution channels, and strong financial resources.

These advantages enable Google not only to continue to invest in large-scale model research and development, but also to rapidly promote AI technology to a large number of consumers and corporate customers by relying on the existing product ecosystem.

Jones anticipates that Google will continue to launch more innovative products around Gemini 4 in the future, including models for different application scenarios and providing different price options to meet the differentiated needs of customers for performance and cost. Based on the above judgment, Jones maintained Alphabet's “outperforming the market” rating, with a target price of $420.

Wall Street expects Google to launch consumer-grade AI smart devices

In addition to improving model performance, Wall Street is also concerned about whether Google will further enter the consumer-grade AI smart device market.

Jones said it is not ruled out that Google will launch a consumer-grade AI smart device similar to Muse under Meta (META.US) in the future and use Gemini technology and its broad consumer product ecosystem to promote popularization.

Unlike traditional chatbots, which mainly provide question and answer and content generation services, AI agents can perform more complex tasks according to user instructions, including retrieving information, planning workflows, and calling external tools to complete operations.

For Google, its huge consumer product ecosystem may provide an important application foundation for AI agents, further extending Gemini's usage scenarios to daily work and life. However, Google has yet to officially announce the launch of consumer-grade smart devices similar to Muse, and the relevant judgment is still part of analysts' expectations for the company's future product direction.