After falling behind in the artificial intelligence race, Google has turned to the Gemini 4 model, introducing a new high-level AI model, Argon. This model is intended to be the foundation of the Gemini 4 generation in an attempt to catch up with competitors Anthropic and OpenAI.
A company representative stated that Argon surpasses Google's previous advanced 'Pro' models in size. He noted that it is their most performant model, created for complex tasks, and compared it to leading models such as OpenAI's Astra and Anthropic's Opus across key metrics in coding and cybersecurity.
Google announced that it is providing this model to select cybersecurity partners and participating in a voluntary Trump administration process for early access to the models. However, public release timelines have not been disclosed.
According to Google's own tests described in the press release, Argon outperformed Astra and Opus on several industry benchmarks, although it lagged behind them in two out of four coding tests included by Google.
Furthermore, it has become known that Google no longer plans to release Gemini 3.5 Pro, which CEO Sundar Pichai had previously announced for June. Meanwhile, Google has reorganized its AI laboratory, DeepMind. Co-founder Demis Hassabis stepped down as CEO in August, becoming chairman of the division, and several Gemini project leaders have left the company.
Lagging Behind
These delays have caused Google to fall behind Anthropic and OpenAI, who have continued to push the boundaries of research with new releases of their flagship models. Although Pichai objected in July to the idea that Google was losing ground in the AI race, in recent months Google has changed its communication strategy, shifting the focus from Gemini's advanced capabilities to cost advantage compared to competing models.