At its annual Google I/O conference, Google introduced significant innovations in artificial intelligence, dominated by the new Gemini 1.5 Flash model. This model is designed to provide high speed and efficiency while maintaining the ability to process vast amounts of data, including text, images, and video files. With this step, Google is responding to the need of developers and businesses for a lighter and more cost-effective solution for common AI tasks.
A key innovation is the deeper integration of artificial intelligence directly into Google Search. The AI Overviews feature (AI-generated summaries) is starting to roll out widely to users. The search engine will no longer just offer a list of links, but will generate a comprehensive answer to the query at the top of the page, summarizing the most important information from various web sources.
Developers are gaining access to an improved version of the Gemini 1.5 Pro model, which now features a context window with a capacity of up to 2 million tokens. In practice, this means that artificial intelligence can analyze, for example, thousand-page documents or hours of video footage at once. This capacity is currently the highest among commercially available AI models on the market.
“Gemini 1.5 Flash is optimized for speed and efficiency where low latency is most important,” company representatives stated during the presentation of new tools for the developer community.
Google also presented future visions in the form of Project Astra, a prototype of a universal AI assistant capable of real-time communication via a smartphone camera. The assistant can identify objects around the user, answer questions about visual cues, or help plan tasks based on what it currently “sees.”
Last but not least, innovations will also affect creative tools. The Veo model will enable the generation of high-resolution videos based on text input, directly competing with similar technologies from companies like OpenAI (Sora) or Adobe. All these innovations underscore Google's strategy to become an “AI-first” platform and integrate generative models into the entire ecosystem of its services.