Google Company presented a new Gemini 2.5 Flash AI model focused on advanced developer capabilities.
This is the first hybrid system that allows you to adjust the level of “thinking” of the model, providing a balance between quality, speed of execution and cost. It is already available in test mode through Gemini API, AI Studio and Vertex AI.
The model is built on the basis of the previous version 2.0 of Flash, keeping its high speed and low cost, but significantly improving the ability of logical analysis. Developers got the opportunity to independently adjust the intensity of calculations and determine the budget for each request, which allows effective management of resources and performance.
Google has released test results that demonstrate the flexibility of Gemini 2.5 Flash in solving tasks related to mathematics, science, logic and software code generation. While the model doesn’t always beat the competition on all counts, it offers a balanced combination of performance, value and accuracy.
The model is also available to users in the Gemini app, where new features can be tested, including Canvas, a space for collaborative work on texts and programming code.
