Accelerating AIGC Adoption: Major Breakthroughs in China's Large Model Development!

According to industry reports, AI startup Moonshot AI has made significant strides in long-context window technology for large models. Their Kimi smart assistant now supports an impressive 2 million words in ultra-long, lossless context and has launched an internal beta test.

Founded by Yang Zhilin, a former student of Apple's AI head and ex-employee of Facebook AI Research, Moonshot AI stands out with Kimi's remarkable ability to manage long texts. The assistant is focused on text processing, information retrieval, and language comprehension.

Recently, Moonshot AI raised over $1 billion in a new funding round, attracting investors such as Alibaba, Sequoia China, Xiaohongshu, and Meituan, elevating its post-investment valuation to approximately $2.5 billion. Many experts believe that 2024 will be a pivotal year for large model applications as these technologies become more stable and widely adopted.

Fang Guangzhao from Open Source Securities notes that advancements in AI multimodal large models, including GPT-4 Turbo, Google Gemini, Google Genie, and Claude 3, are progressing rapidly. Improved performance combined with declining costs is expected to accelerate the adoption of AI-generated content (AIGC) applications.

Data from the Financial Association indicates that Yidian Tianxia's AIGC product, KreadoAI, features a range of multimodal capabilities. These include text generation, image synthesis, video creation from text, speech generation, voice cloning, and the creation of digital humans. Similarly, Fantasia Technology is leveraging AIGC technology across text, images, and virtual personas, collaborating with Baidu to develop Art Goose, the National Grand Theatre's first virtual employee.