On Christmas Eve 2024, OpenAI's 12 days of live streaming came to an end, but Google's breakthrough in the field of AI stole the show. Since the advent of ChatGPT, Google has accelerated its catch-up and recently moved frequently.

In December, Google launched its more advanced AI model Gemini 2.0 series, which supports multi-modal input and output, surpassing its predecessors in a number of tests. At the same time, Google also released three prototype AI agents based on Gemini 2.0, as well as an upgraded version of Project Astra, which can convert languages in real-time in multilingual environments.

In terms of inference model, Google launched Gemini 2.0 Flash Thinking, which is benchmarked against OpenAI o1 and o3. In video and image generation, Google's Veo 2 and Imagen 3 are compared to Sora and DALL-E 3, respectively, where Veo 2 is superior in resolution and video length.

In addition, Google also introduced a new experimental image mashup tool Whisk, as well as a sixth-generation tensor processing unit Trillium and a quantum chip Willow. The founder of DeepMind has won the Nobel Prize in Chemistry for developing AlphaFold, an AI system that successfully predicts the structure of proteins.

In contrast, OpenAI showed off most of its previously demonstrated features at the live event, such as the ChatGPT search engine and Siri integration. The Financial Times pointed out that Google's recent AI development has been highly praised, instilling confidence in investors.

(Note: The article comes from the network, the information is for reference only, does not represent the views of this website, if there is infringement, please contact to delete!)