Google has officially launched its highly anticipated new flagship AI model, Gemini 2.0 Pro Experimental. This announcement was made as part of a series of presentations on other AI models. Additionally, Google plans to introduce its “logical reasoning” model, Gemini 2.0 Flash Thinking, to the public through the Gemini app.
However, while Google showcases such AI models, the tech world users are paying significant attention to the more affordable logical reasoning models from the Chinese startup DeepSeek. DeepSeek’s models are reportedly capable of competing with, and even surpassing, advanced AI models created by American technology giants. Companies can also rent these models at a relatively low cost using the DeepSeek API.
Although Google and DeepSeek both released AI-based “reasoning” models simultaneously last December, DeepSeek’s R1 model gained more attention. Now, it is evident that Google aims to reach a wider audience with its popular Gemini app by introducing the Gemini 2.0 Flash Thinking model.
Google announced Gemini 2.0 Pro as a successor to the Gemini 1.5 Pro model and praised it as the leading example in the Gemini AI model family.
A week ago, the company unexpectedly announced the release of Gemini 2.0 Pro in the change log of the Gemini app. This time, however, all the information has been officially confirmed. Google plans to launch a test version of this model on the Vertex AI and Google AI Studio platforms. Additionally, Gemini app subscribers will have access to Gemini 2.0 Pro.
According to Google, the new model is highly efficient in coding and performing complex tasks and offers an improved “understanding and logical analysis of world knowledge” compared to previous models. Gemini 2.0 Pro can also interact with tools like Google Search on behalf of users and can write code or give commands when necessary.
Furthermore, the context window of Gemini 2.0 Pro consists of 2 million tokens, allowing it to process approximately 1.5 million words at once. In other words, the latest model from Google can fully encompass all seven books of “Harry Potter” in a single query and still have a “reserve” capacity for about 400,000 more words.
Google is making the Gemini 2.0 Flash model available to the public. This model was announced in December 2024 and will now be open to all Gemini app users.
In summary, Google is announcing the new, more economically efficient Gemini 2.0 Flash-Lite model to counteract or create competition with the increasing interest in DeepSeek. The company has stated that this model will deliver higher and more accurate results compared to Gemini 1.5 Flash but will operate at the same price and speed.
Prepared by Zukhrakhon Mansurova
Leave a Reply