
Earlier, Google released a preview version of Gemini 2.5 Pro to developers, encouraging them to explore its capabilities and provide feedback. Now, the official release of Gemini 2.5 Pro has arrived—accompanied by the introduction of the Gemini 2.5 Flash-Lite edition.
The Flash-Lite variant outperforms its Gemini 2.0 predecessors in coding, mathematics, and reasoning benchmarks. Designed primarily for handling large-scale tasks such as translation and classification, it strikes a balance between performance, affordability, and speed.
Tasks like translation and classification typically involve massive volumes of data, and some demand swift AI processing. Flash-Lite excels in these scenarios by offering commendable performance at a remarkably low cost, with faster response times compared to its peers.
Flash-Lite is priced at just $0.10 per million input tokens and $0.40 per million output tokens—a rate that can only be described as exceptionally economical. However, its capabilities in complex reasoning tasks are limited, making it ill-suited for more demanding use cases.
As for Gemini 2.5 Pro, it requires little elaboration—having already demonstrated impressive performance across a range of benchmark tests, it is now ready for deployment in enterprise production environments.
Developers can begin experimenting with both models immediately. Google has made Gemini 2.5 Pro available through Google AI Studio, and today’s update also adds access to the Flash-Lite version.
However, at present, developers cannot access the Gemini 2.5 series APIs for free. While Google has previously stated that a free tier will be offered, it may come with certain limitations. In response to developer frustration over AI Studio’s lack of free access to Gemini 2.5 Pro, Google remarked: A free version may be provided in the future, albeit with some restrictions.