Google is preparing for a significant expansion of its AI capabilities with the upcoming general release of Gemini 2.5 Pro in June 2025.
According to recent announcements from Google DeepMind, Gemini 2.5 Flash is now available to everyone in the Gemini app, with plans to make it generally available in Google AI Studio for developers and in Vertex AI for enterprises in early June. The more powerful Gemini 2.5 Pro version will follow shortly after.
The standout feature of the 2.5 Pro model is Deep Think, an experimental enhanced reasoning mode that allows the model to consider multiple hypotheses before responding. This capability is specifically designed for highly complex math and coding tasks, with Google reporting impressive scores on difficult benchmarks like the 2025 USA Mathematical Olympiad and LiveCodeBench, a competition-level coding benchmark.
"Through exploring the frontiers of Gemini's thinking capabilities, we're starting to test an enhanced reasoning mode called Deep Think that uses new research techniques enabling the model to consider multiple hypotheses before responding," Google stated in a recent blog post.
Security has been a major focus for the Gemini 2.5 family. Google has implemented advanced security safeguards that have substantially increased Gemini's protection against indirect prompt injection attacks during tool use. These attacks occur when malicious instructions are embedded into data that an AI model retrieves. Through automated red teaming and model hardening, Google has significantly reduced the success rate of such attacks, making the 2.5 family their most secure model to date.
Beyond security improvements, Google is bringing Project Mariner's computer use capabilities into the Gemini API and Vertex AI. Companies including Automation Anywhere, UiPath, Browserbase, and others are already exploring its potential for automation and workflow enhancement.
Additional features coming to both 2.5 Pro and 2.5 Flash include thought summaries, which organize the model's reasoning process into a clear format, and native audio capabilities with support for multiple speakers across 24+ languages.
As Google continues to compete in the increasingly competitive AI landscape, these enhancements to Gemini 2.5 Pro represent significant advancements in reasoning capabilities, security, and multimodal functionality.