Google announced the official release of the Gemini 2.5 Pro and Gemini 2.5 Flash models from preview status. They have become stable for use in Google AI Studio, Vertex AI, the Gemini app, and search. Additionally, the company launched a preview of the Gemini 2.5 Flash-Lite model, designed for tasks with large data volumes and minimal latency, capable of responding in less than 100 milliseconds at a lower cost per token.
Gemini 2.5 Pro retains context up to one million tokens and offers enhanced management for developers. It has received the “Deep Think” feature, allowing the model to consider multiple solutions for complex mathematical or programming tasks. Audio output support, computer actions through Project Mariner, and additional protection against indirect prompt injections have also been introduced.
Gemini 2.5 Flash has become stable and is available to developers with updated pricing: $0.30 per million input tokens and $2.50 per million output tokens. The price difference between thinking and regular processing modes has been eliminated for this model, and a single pricing tier has been introduced regardless of input data size. Flash-Lite is focused on translation and classification tasks where speed and low cost are important.
Users of the free version of the Gemini app have limited access to 2.5 Pro, while AI Pro subscribers can send up to 100 requests per day, and Google AI Ultra users receive the highest level of access.
