Google recently launched three new Gemini 2.0 models, aiming to provide diverse performance and cost options for different application scenarios.
Gemini 2.0 Flash Thinking
The model, which focuses on inference capabilities, is now available in the Gemini app on desktop and mobile devices and can be easily selected through the model drop-down menu. The Flash Thinking Experimental with apps feature can also watch YouTube videos and extract key information to enhance the user experience.
Gemini 2.0 Flash-Lite
This is an affordable version for developers and is now available in public preview via the API, giving developers more options.
Gemini 2.0 Pro Experimental
Designed for complex prompting and encoding tasks, this model has an extended context window of 2 million tokens, twice as much as the Flash version, demonstrating powerful processing capabilities for users who need high performance.
Currently, these models only support text output, but they all have the ability to process images and audio as input, which is expected to bring more application possibilities in the future.