Technology giant Google has unveiled a wide array of new products and updates at its annual developer conference, Google I/O 2025. Highlights include groundbreaking advancements such as 3D video calls via Beam, cutting-edge image and video generation models like Imagen 4 and Veo 3, and introduction of Android XR, a dedicated platform for smart wearables. Sundar Pichai, CEO of Google and Alphabet, emphasised the accelerating global adoption of artificial intelligence, stating: “The world is responding, adopting AI faster than ever before.” He also shared that the Gemini app now boasts over 400 million monthly active users, with usage of the Gemini 2.5 Pro model increasing by 45%.
Gemini 2.5 Enhancements
Google introduced substantial upgrades to its Gemini 2.5 Pro and Flash models. Gemini 2.5 Pro received an experimental “Deep Think Mode” for enhanced reasoning, allowing the AI to explore multiple hypotheses. Both Pro and Flash models gained native audio outputs, enabling them to generate expressive, human-like speech. Gemini 2.5 Flash also saw improvements in efficiency and security against prompt injection attacks, signifying a leap in Gemini’s capabilities for complex queries and multimodal interactions.
A transformative overhaul of Google Search was announced, integrating an “AI Mode” that facilitates chatbot-style conversations and follow-up questions for more conversational experiences. This new mode, powered by a custom Gemini 2.5 model, is rolling out to US users. Key features like “Deep Search” for comprehensive research reports and “Search Live” for real-time visual inquiries using a device’s camera aim to redefine how users interact with information.
Google introduced two new premium AI subscription tiers: Google AI Pro ($19.99/month) and Google AI Ultra ($249.99/month). Google AI Pro offers higher usage limits for the Gemini app and access to tools like Flow and NotebookLM. Google AI Ultra, the premium offering, provides the highest usage limits, early access to cutting-edge models and features like Veo 3 and Agent Mode, 30 TB of storage, and YouTube Premium, marking Google’s strategic move toward monetizing its advanced AI capabilities.
Google Beam, a next-generation 3D video conferencing system, moved from research to commercialization. Designed to replicate in-person interactions, Beam uses AI, advanced 3D imaging, and specialized displays to create life-size, high-fidelity images of participants without requiring headsets. In collaboration with HP, the first Google Beam devices are set to be available for early customers later this year, with planned compatibility for Google Meet and Zoom.
Google unveiled Android XR, a new operating system specifically designed for augmented, mixed, and virtual reality devices. This platform integrates Gemini AI directly into headsets and smart glasses, enabling hands-free AI assistance for real-time translation, messaging, and navigation. Partnerships with major eyewear brands like Samsung (Project Moohan), Gentle Monster, and Warby Parker were also announced, signaling a strong push into wearable computing.
Significant advancements in generative AI for media were showcased with Imagen 4 and Veo 3. Imagen 4, Google’s most advanced image generation model, offers improved clarity, fine detail rendering, better spelling and typography, and support for 2K resolution. Veo 3, the latest video generation model, can now produce not only realistic visuals from text prompts but also accompanying audio, including ambient sounds and character dialogue, enhancing the immersive quality of AI-generated content.
Google introduced Flow, a new AI filmmaking tool built with and for creatives. Leveraging Google DeepMind’s models, Flow allows users to weave cinematic films with precise control over characters, scenes, and styles. It enables the creation and stitching together of AI-generated video clips, offering a powerful new avenue for content creation, available to Google AI Pro and Ultra subscribers.
Gemini Live, the conversational AI with vision capabilities, is now globally available for free on Android and iPhone. This feature enables Gemini to interpret objects through a device’s camera and respond contextually, providing real-time assistance and expanding how users can interact with AI in their daily lives.
Gemini is becoming more personalised with ‘Personal Context’ using a user’s search history for proactive, tailored assistance. Additionally, ‘Canvas’ allows for co-creation of infographics, quizzes, and podcast-style Audio Overviews in multiple languages, making Gemini a more versatile tool for productivity and creative tasks.
To address growing concerns about AI-generated content, Google launched the SynthID Detector, a verification portal designed to identify AI-generated content watermarked with SynthID. This tool aims to provide transparency by allowing users to upload media and check if it was created using Google’s AI models, bolstering efforts to combat misinformation and promote responsible AI use.