🔎 Google I/O 2025: A Bunch of AI Innovations from Google!
Today, Google started its annual Google I/O conference. Already on day one, the corporation announced a lot of updates to its AI services and the launch of new ones.
What's New?
➡️ Gemini 2.5 Flash update. The model codes better, better handles images and long contexts, all while consuming up to 30% fewer tokens.
Available now for free on AI Studio.
➡️ Veo 3 video generator. The model creates videos with sound. It will be available on the AI Ultra plan for $250 per month.
➡️ Deep Think mode for Gemini 2.5 Pro. It surpasses OpenAI o3, developers claim.
➡️ Gemini Diffusion. A language model based on a paradigm familiar to image generators: it doesn't just predict the next text token, but creates the entire response from "white noise."
➡️ NotebookLM now creates video summaries based on your texts, PDFs, websites, and other sources.
➡️ Jules. An AI agent for cloud programming — a competitor to OpenAI Codex — is now available to everyone.
➡️ Google Meet now translates conversations in real time.
➡️ Project Astra. An update to the voice assistant introduced last year — it can control your device directly during a conversation and remembers your past conversations.
And these are just the main things. Google is also opening up its AI Mode in Search for everyone in the U.S. and expanding AI model management capabilities for developers.
Let's see what they'll show tomorrow on the second day of the conference. And, most interestingly, how will competitors react?
