OpenAI just announced their new flagship multi-modal GPT-o ('o' stands for Omni) which can reason across audio, video, and text in real-time. This is a historically significant step in natural human-computer interaction. GPT-o uses a single end-to-end model to process text, audio, and video, improving latency and the natural understanding/generation of expressions and emotions. The most impressive thing I noticed is the voice, which uses appropriate feelings and tonal language during interactions. GPT-o will be free to use and will roll out to all users in the coming weeks, with voice interaction available only to paid users.

AlphaFold 3, a new AI model by Google DeepMind, can now predict the structure in all of life’s proteins, DNA, RNA, and their interactions. Protein folding has been a challenging problem for ages due to its computational complexity, and these AI-based approaches are significant examples of how AI can benefit science directly.

Google announced their generative video model, Voe, at the recent Google I/O keynote. It can generate 1080p resolution videos that can exceed one minute in length. This model is expected to be in similar caliber to sora, OpenAI's state-of-the-art video generation model. They are also planning to integrate the video generation capability directly into YouTube for creators.

React team just open-sourced the react compiler which has been in development for past 2 years. It can be used with react 19 (still experimental) to improves performance of react codebases by reducing re-rendering in components. No more manual useMemo and useCallback in the code to improve performance.

Experience GPT-4o capabilities in latest OpenAI spring update event.

Google announced wide range of AI products in their latest Google I/O keynote

Apple introduced their latest SoC (System on a Chip) with up to 1.5x faster CPU performance over M2 and x4 faster performance in GPU over M2.

Testing email for multiple clients is can be a pain (specially in outlook). Caniemail can used to check supported features vs clients easily.