Workflow
音频人工智能模型
icon
Search documents
全球大公司要闻 | 宇树科技:未涉及申请“绿色通道”相关事宜,上市工作正常推进
Wind万得· 2026-01-04 22:34
Group 1 - TSMC has received an annual license from the US government to export chip manufacturing equipment to its factory in Nanjing, ensuring uninterrupted operations and product delivery. The 2nm process mass production plan is on schedule, with risk trial production expected to start in 2027 [2] - Guizhou Moutai has spent 120 million yuan to repurchase approximately 87,100 shares of its stock. Additionally, the maximum purchase limit for its Feitian 53% vol 500ml Moutai liquor has been adjusted from 12 bottles to 6 bottles per person per day from January 4 until before the Spring Festival [2] - Xiaomi's CEO Lei Jun revealed that the company aims to deliver over 410,000 vehicles in 2025 and 550,000 in 2026, addressing concerns about the materials used in their vehicles and refuting rumors about the company's marketing practices [3] Group 2 - Baidu's AI chip subsidiary Kunlun has officially submitted its main board listing application to the Hong Kong Stock Exchange, with its valuation increasing from approximately 13 billion yuan in 2021 to 21 billion yuan by July 2025 [5] - BYD's pure electric vehicle sales are projected to reach 2.2567 million units by 2025, surpassing Tesla's expected delivery of 1.636 million units, marking the first time BYD has outperformed Tesla in annual sales [5] - SK Hynix's target price has been raised to 840,000 KRW by Morgan Stanley, indicating a significant improvement in the company's profitability trajectory [11]
OpenAI加码音频人工智能研发 备战首款硬件设备
Xin Lang Cai Jing· 2026-01-01 15:39
Core Insights - OpenAI is preparing to upgrade its audio AI model to launch its first AI-driven personal hardware device, focusing on audio interaction as a core feature [1] - The audio version of ChatGPT will allow users to interact via voice, but the underlying large language model for audio is different from the one used for text interactions [1] - OpenAI researchers have identified that the current audio model lags behind the text model in terms of response accuracy and speed, prompting a consolidation of multiple teams to optimize the audio model for future hardware [1]