豆包视频生成模型Seedance 1.0 Pro

Search documents
张一鸣久违现身,大佬纷纷杀回一线
Hua Er Jie Jian Wen· 2025-10-11 09:12
作者 | 王小娟 编辑 | 黄昱 张一鸣的这次公开亮相,意义远超过一场普通的揭牌仪式。这是他自2021年卸任字节跳动CEO四年来首 次公开露面,引发了业内广泛关注。 久违的,又一位互联网大佬重新回归大众视野。 在2021年卸任后,张一鸣一度远离聚光灯。当时,他在内部信中表示:"我不是传统意义上的成熟管理 者,我也不是很擅长社交,我更喜欢研究组织和市场原理来减少管理,喜欢自己上网、看书、听歌、发 呆。" 10月9日,上海徐汇知春创新中心正式揭牌,字节跳动创始人张一鸣出席启动仪式并发表讲话。该机构 定位为民办非企业单位,由张一鸣与上海交通大学ACM班创始人俞勇教授联合创办,专注前沿计算机 和人工智能技术创新研究、开源工具和算法开发、以及顶尖创新人才培养。 而卸任也不代表着完全甩手。当时,他也表示,自己将"相对专注学习知识,系统思考,研究新事物, 动手尝试和体验,以十年为期,为公司创造更多可能"。 这位于2021年卸任CEO的互联网领军人物,在沉寂四年后以全新姿态回归公众视野。在这里,他主要谈 到对人才培养的一些看法。 如今,AI作为新的变量,成为各大互联网公司被重估的重要考量,张一鸣信中"公司的更多可能",也与 ...
【产业互联网周报】字节AI Lab负责人李航被曝正式卸任;OpenAI ARR破100亿美元;谷歌启动全公司范围“自愿离职计划”
Tai Mei Ti A P P· 2025-06-16 08:33
Financial Performance - Oracle reported a total revenue of $57.4 billion for the fiscal year 2025, with a net profit of $12.4 billion. The fourth quarter revenue reached $15.9 billion, marking an 11% year-over-year increase [2] - Adobe's Q2 revenue was $5.87 billion, up 11% year-over-year, exceeding analyst expectations. Adjusted EPS was $5.06, a 13% increase, but investor concerns about future competition in AI remain [3] Company Changes - ByteDance's AI Lab head, Li Hang, has officially stepped down, transitioning to a consultant role. This change follows the integration of the AI Lab into the Seed team led by former Google DeepMind VP Wu Yonghui [3][5] - Kingsoft Cloud announced the resignation of CFO He Haijian, effective June 30, 2025, due to personal reasons [6] AI Talent Acquisition - Baidu launched its largest AI talent recruitment initiative, expanding job openings by over 60% compared to last year, focusing on cutting-edge AI fields [6] AI Product Developments - ByteDance's self-developed AI IDE "TRAE" has surpassed 1 million monthly active users, with over 80% of internal engineers utilizing it for development [7] - Volcano Engine released version 1.6 of its Doubao large model, introducing a pricing model based on input length, significantly reducing costs [9] - Alibaba's Tongyi Qianwen 3 model achieved over 12.5 million downloads in its first month, with over 130,000 derivative models created [10] Corporate Investments - Amazon plans to invest AUD 20 billion in expanding its data center infrastructure in Australia from 2025 to 2029, marking the largest global tech investment in the country [18] - Meta announced the acquisition of a 49% stake in Scale AI for $14.8 billion to enhance its AI capabilities [22][23] Market Trends - IDC forecasts that the Chinese graph database market will reach RMB 644 million in 2024, growing 17% year-over-year, with major players including Huawei Cloud and Ant Group [24] - The Ministry of Transport emphasized the need for collaboration in technology transfer within the transportation sector to enhance AI applications [27]
使用成本降至三分之一字节大模型战略升级
Zheng Quan Shi Bao· 2025-06-11 17:25
Core Insights - The article highlights the significant upgrade of the Doubao large model and the introduction of an innovative "interval pricing" model, marking a "value war" to balance cost and performance [1][3] Group 1: Product Launch and Features - On June 11, ByteDance's Volcano Engine held the Force Original Power Conference, where the Doubao large model family was fully upgraded, including the release of Doubao Model 1.6, Seedance 1.0 Pro, and new AI cloud-native services [1] - Doubao Model 1.6 is a comprehensive model supporting 256K long context and adaptive reasoning, with enhanced capabilities in deep reasoning, coding, mathematics, and logic [1] - Doubao Model 1.6—thinking excels in complex reasoning and has ranked among the top globally in various authoritative assessments [1] Group 2: Pricing Strategy - The pricing strategy for Doubao Model 1.6 features two core aspects: unified pricing to eliminate functional premiums and segmented scheduling to match demand accurately [2] - The token price remains consistent regardless of whether deep reasoning or multimodal functions are activated, allowing companies to avoid unnecessary costs [2] - The pricing intervals are categorized based on input length, with the most used 0-32K input range priced at 0.8 yuan per million tokens for input and 8 yuan per million tokens for output, making it one-third the cost of previous models [2]