AI 101: Google Launches Groundbreaking AI Models, Taking on OpenAI
How Google is challenging OpenAI’s market dominance
Welcome to another edition of AI 101, where every Wednesday we bring you the biggest AI update of the week.
This week’s update includes contributions from our editorial team: Melinda Mei and Luna Meline.
This Week’s Update: Google Releases Gemini 2.0 and Veo 2
Last Wednesday, Google DeepMind launched Gemini 2.0. The new model introduces multimodal capabilities, supporting inputs like images, video, and audio while generating outputs such as text-to-speech audio and mixed text-image content—something OpenAI’s most sophisticated models, GPT-4o and o1, cannot do. Gemini 2.0 Flash Experimental, the first model in the 2.0 family, outperforms Google’s last flagship model on almost every benchmark at twice the speed. Early benchmark scores also rank 2.0 as one of the top AI models globally, achieving comparable quality to GPT-4o and o1 at a higher speed and lower cost.
Google also unveiled Veo 2, a groundbreaking AI video-generation model capable of producing 4K clips over two minutes long. The new model features an enhanced understanding of physics, camera control, and nuanced human expression, creating clearer footage. Currently, the model is available on a waitlisted basis with videos capped at eight seconds and 720p resolution.
Why This Is Important
The launch of Google’s new models represents a significant challenge to OpenAI’s dominance in the AI industry. Google’s Veo 2 video-generation model is outperforming OpenAI’s Sora model which was released earlier this month. Users have also rated Gemini 2.0 highly, landing it at number three on the LLM leaderboard, a site where users blindly score different generative AI models. Gemini 2.0 Flash is only the first model in the 2.0 family; later models will likely offer even better performance. As competition rises, OpenAI faces increasing pressure to innovate and release new products or risk losing market share.
Quick Hits:
Last week, authorities confirmed that former OpenAI researcher and whistleblower Suchir Balaji was found dead in his San Francisco apartment. He had raised concerns about potential violations of copyright law by OpenAI earlier this year.
ChatGPT experienced a three-hour outage last week, which OpenAI attributed to the installation of a new monitoring tool that backfired.