Google Employees Question Gemini 4 Argon AI Model's Performance

Google's long-awaited Gemini 4 Argon AI model has been rolled out to employees, but internal doubts remain about its performance. The model is reported to struggle with certain coding tasks, despite performing well on industry benchmarks.

According to Bloomberg, some Google employees believe that OpenAI's Astra and Anthropic's Fable models are improving at a faster rate than Gemini 4. This has led to internal skepticism over the model's ability to compete in key areas such as coding.

In response to these concerns, Koray Kavukcuoglu, head of Google DeepMind, stated that he is 'encouraged' by the model's performance and has 'the utmost trust in the team.'