Google launches Gemini 2.5 Flash, and this is what it can do

Google has released an early preview of its new Gemini 2.5 Flash model, now available through the Gemini API in Google AI Studio and Vertex AI. Building on the success of Gemini 2.0 Flash, the new model offers major improvements in reasoning ability while continuing to prioritize speed and cost-efficiency.

Gemini 2.5 Flash introduces a hybrid reasoning system, allowing developers to switch “thinking” modes on or off and adjust thinking budgets—a new way to fine-tune the balance between quality, cost, and latency. Even with reasoning turned off, the model performs faster than its predecessor, while still delivering better results.

Also read: Google Launches Veo 2, text-to-video AI generator

According to Google, the new model achieves an excellent performance-to-cost ratio, making it one of the most efficient AI tools currently available. It’s designed to help developers build faster and smarter applications with flexible control over how the AI thinks and processes data.

The model is also integrated into the Gemini app, where users can explore features like Canvas—an interactive environment for editing documents and code with AI assistance.

Google encourages developers and users to test Gemini 2.5 Flash and share feedback, as it prepares for a broader rollout.