In a surprising shift from tradition, Google opted not to introduce its latest flagship AI model at the annual I/O conference. CEO Sundar Pichai’s decision to postpone the release of the Gemini 3.5 Pro AI model has sparked significant discussion about Google’s strategic positioning in the competitive AI coding landscape.
During the keynote, Pichai revealed that the Gemini 3.5 Pro was still in development, eliciting audible disappointment from the audience. While some speculate on the reasons for this delay, a plausible theory is that Google aims to enhance the model’s proficiency in AI coding tasks before its official launch.
In the rapidly evolving field of AI, platforms like Anthropic’s Claude Code and OpenAI’s Codex have set high standards by advancing AI coding capabilities. These innovations have captured the attention of developers and generated substantial revenues by streamlining and automating coding processes.
Instead of unveiling the 3.5 Pro, Pichai focused on introducing Gemini 3.5 Flash—a smaller, faster, and more cost-effective model. Despite being slightly less powerful than leading models, 3.5 Flash is now the primary engine behind Google’s Antigravity AI coding service, available to developers starting today.
This strategic move allows Google to gather extensive and valuable anonymous data from developers using Antigravity. For example, if a project initiated in Antigravity is suddenly halted, it indicates potential issues with the Flash 3.5 outputs. This feedback is instrumental in refining the 3.5 Pro model through reinforcement learning, a method that sharpens AI by reinforcing successful outputs and addressing failures.
The data and insights from the 3.5 Flash model are expected to play a crucial role in enhancing the capabilities of the larger 3.5 Pro model. The clarity and precision inherent in coding tasks provide a solid foundation for AI model improvement, as successful code tends to function correctly, while erroneous code typically leads to failures.
Onstage, Pichai expressed anticipation for the upcoming release, stating, “I know you can’t wait to get your hands on it. Give us until next month to get it to you.”
As the tech community eagerly awaits the launch of the Gemini 3.5 Pro model, expectations are high for its potential to revolutionize the domain of AI-assisted coding, currently one of the most transformative applications of generative AI.
