On January 15, 2025, MiniMax announced the open sourcing of its new series of models, MiniMax-01, which includes the foundational language model MiniMax-Text-01 and the visual multimodal model MiniMax-VL-01. The MiniMax-01 series features bold innovations in its architecture, implementing linear attention mechanisms on a large scale for the first time, breaking the limitations of traditional Transformer architectures. With a parameter count reaching 456 billion and 45.9 billion activations per instance, its overall performance competes with international standards.

During the first session of the 2024 Ideal AI Talk, Ideal Auto CEO Li Xiang shared his views on the future role of artificial intelligence in the automotive industry and announced that the company will shift towards becoming an AI enterprise. Li believes that intelligence is the future of the automotive industry, and Ideal Auto will play an important role in this transformation. In terms of R&D investment, Li revealed that half of the 10 billion yuan budget allocated for R&D each year is dedicated to the field of artificial intelligence. He believes that the emergence of large models will bring fundamental changes to humanity, just as the internet brought about information revolution.

Chinese AI video generation startup Aishi Technology recently announced the completion of its Series A2 to A4 financing, raising nearly 300 million yuan. Previously, the Series A2 financing was funded by Ant Group, while the recent Series A3 and A4 rounds were co-invested by the Beijing Artificial Intelligence Industry Investment Fund, Guoke Investment, and Guangyuan Capital. This round of financing will primarily be used to enhance technological research and development capabilities, expand computing power resources, build a talent team, and accelerate product feature iteration and market coverage, thereby continuously promoting the widespread application of AI video generation technology.

Canadian AI hardware startup Tenstorrent has recently announced that it successfully completed a funding round of nearly $700 million. This funding round totaled $693 million, bringing Tenstorrent's valuation to over $2.6 billion. The round was led by Samsung Securities and AFW Partners, with other investors including Hyundai Motor and Jeff Bezos's investment company Bezos Expeditions. Image source note: The image is AI-generated, licensed from Midjourney.

Video editing platform InVideo recently launched a significant upgrade version 3.0, introducing AI-based text-to-video generation capabilities. Users can create real-life footage, animated, or anime-style video content simply by entering text prompts, with customizable adjustments for different platforms such as YouTube, Instagram Reels, and LinkedIn. Although this new feature relies on third-party AI models, users can dynamically modify the video in real-time by adding text prompts.

American semiconductor company AMD has recently announced a global workforce reduction of about 4%, with the number of layoffs expected to be under 1,000. This move comes as AMD continues to surpass Intel in the x86 processor market but faces fierce competition from industry giant Nvidia in its transition from graphics processing units (GPU) to artificial intelligence (AI) accelerators. An AMD spokesperson stated, 'To align our resources with our biggest growth opportunities, we are taking some targeted steps.'

According to reports, Microsoft will spend nearly $10 billion renting servers from AI startup CoreWeave between 2023 and 2030. This amount accounts for more than half of the total $17 billion contract signed between CoreWeave and its customers. Neither CoreWeave nor Microsoft responded immediately to requests for comment. Previously, it was reported that CoreWeave is preparing for an initial public offering (IPO) in 2025. The company recently secured a $650 million credit line from several investment banks.

Is AI actually starting to ‘get smarter’?! The latest research from MIT indicates that the internal structure of large language models (LLMs) astonishingly resembles that of the human brain! This study utilized sparse autoencoder techniques to conduct an in-depth analysis of the activation space of LLMs, revealing three hierarchical structural features that are truly remarkable: first, at the microscopic level, researchers discovered the existence of structures akin to ‘crystals’. The surfaces of these ‘crystals’ are composed of parallelograms or trapezoids, similar to familiar lexical analogies such as ‘man:’

Google is expanding the 'Help me write' feature to the web version of Gmail, allowing users to craft or refine emails using Gemini AI. Similar to the mobile version, users will see prompts to use this feature when opening a blank draft in Gmail. However, it is important to note that the 'Help me write' feature is not available to everyone; it is only accessible to subscribers of Google One AI Premium or those with Gemini W.

Recently, a research team from Princeton University released an intriguing report indicating that as of August 2024, approximately 4.36% of new articles on Wikipedia contain significant AI-generated content. This study was conducted by scholars Creston Brooks, Samuel Eggert, and Denis Peskoff, who used tools named GPTZero and Binoculars to detect this AI-generated content.