Google unveiled the second generation of AI models in the Gemini mobile app

Google has announced an update to its AI models for the Gemini platform, introducing experimental versions of Gemini 2.0 Pro and Flash-Lite. These developments, the company claims, are the “most advanced” in its arsenal. Gemini 2.0 Pro stands out with improved analytics, complex query processing and code generation capabilities, as well as context window support for up to 2 million tokens, which allows the model to handle huge amounts of data. The model is already available to Gemini Advanced subscribers.

.

Available and cost-effective

In addition, Google announced public test access to Gemini 2.0 Flash-Lite, a simplified version designed to save resources and lightning-fast processing. The model is built into the Google AI Studio and Vertex AI ecosystem, making it easy for developers to use. Both Pro and Flash-Lite are now supported by the Gemini mobile app, giving users the freedom to choose between experimental modes: powerful analytics for complex tasks or fast responses for everyday queries.

Smartphone Innovations

The main new feature is the Gemini 2.0 Flash Thinking Experimental, which not only analyzes data from YouTube, Google Search and Maps, but also combines it to offer answers based on the latest information, for example, a request for a travel itinerary can be supplemented with up-to-date video reviews of places of interest. Another key feature is the “reasoning” mode, which shows in real time how the model breaks down a query into logical steps to form a final answer. This makes the process transparent, speeds up interaction and supports multimodal input: text, images and context for up to 1 million tokens. At the same time, the model’s knowledge is up-to-date as of June 2024, which guarantees the relevance of the data.

Model knowledge is up-to-date as of June 2024, which guarantees the relevance of the data.

Competition and prospects

.

Google’s new model launches come amid stiff competition from the likes of DeepSeek AI and ChatGPT-4o. To strengthen its position, the company has bet on Gemini 2.0 Pro multimodality, which combines text, image, and search data analysis to dramatically improve contextual understanding. For example, when solving a complex problem, the model can match graphs from research with the latest scholarly articles found in Google Scholar. This update not only strengthens Google’s technological leadership, but also expands the scenarios of AI use: from everyday issues like budget planning to professional tasks like big data analysis or code generation. In this way, Gemini is gradually transforming from a tool to a universal digital assistant that can adapt to any need.