At around 1 am today, NVIDIA's stock price rose 3.51% to a new all-time high of $135.58, reaching a market value of $3.34 trillion, surpassing Microsoft and Apple to become the company with the highest market value globally.

Xiaohongshu quietly launched an AI chatbot named "Da Vinci," which can provide intelligent question-and-answer services, including travel guides, food guides, geographical and cultural knowledge, life skills, and more. Some users on Jike APP commented that "the feedback on many details feels very powerful."

What other hot topics in the AI industry at home and abroad are worth paying attention to in the past day? Let's take a look together with Raven Jun.

/ 01 / Large Models

1) Meta Releases Multiple Models: Multi-modal Model Chameleon, Text-to-Music Model JASCO, Audio Watermarking Technology AudioSeal

Meta recently released several research achievements, including the multi-modal model Chameleon, text-to-music model JASCO, audio watermarking technology AudioSeal, etc., bringing new technological breakthroughs and application prospects to the AI field.

2) ChatGPT Experiences Major Outages Again After Two Weeks, Unable to Respond to User Queries

Recently, ChatGPT from OpenAI has been experiencing outages, with the mobile and web versions unable to respond to user queries at times. OpenAI also experienced similar outages two weeks ago and in November 2023. CEO Sam Altman has publicly stated that due to the unexpected usage of new features, there may be service instability.

3) Bilibili Opens Source Lightweight AI Language Model Index-1.9B

Bilibili's newly open-sourced Index-1.9B model has attracted widespread attention. The model includes a base model, a control group, and a dialogue model, with 1.9 billion non-word embedding parameters and leading performance on multiple evaluation benchmarks.

4) Snap Showcases Real-time Device-side Image Diffusion Model at Augmented World Expo

Snap showcased an early version of its real-time device-side image diffusion model at the Augmented World Expo, designed for AR creators with generative AI tools. The model is small and fast, capable of real-time frame re-rendering.

5) L4GM: Generates 4D Gaussian Models for Video Target Objects in Seconds

The research team proposed a large-scale 4D Gaussian reconstruction model called L4GM, which can generate animated objects from single-view video input, achieving impressive results. The model relies on innovative datasets and simplified designs, completing one-way passes in a short time to ensure high-quality output.

6) Hedra's Character-1 Opens for Use, Generating Singing Videos Through Text and Images

Hedra's Character-1 is now open for use, a tool that can generate speaking and singing videos through text and images, supporting a free experience of up to 60 seconds. It is also a brand-new creative platform, providing everyone with unlimited opportunities for video creation. In the future, they will also launch the Worlds feature, allowing users to build their own virtual worlds. (Official website: https://www.hedra.com/)

/ 02 / AI Applications

1) Meta Reorganizes Its Metaverse Team to Focus on MR/AR/AI Features.

Meta CTO announced the largest reorganization of the hardware department since it was renamed Reality Labs in 2020.

According to internal memos to employees: All teams at Reality Labs will be