此语言仍有 24 条内容或界面任务待翻译;未完成部分显示中文原文。

Insights & News

AI、算力、具身智能机器人与智能硬件的产业动态,以及选型与采购指南。

今日更新0 篇Updates This Week23 篇Published147 篇订阅周报

2026-10-22 · 本站更新

具身智能机器人Industry Updates

2026AGIBOT Innovation Day set for 10 29 at Shanghai World Expo Center, focusing on humanoid robot deployment state

According to Zi牛 News, 2026AGIBOT Innovation Day will be held on 10 29 at the Shanghai World Expo Center. AGIBOT has previewed announcements around the deployment state covering hardware embodiment, locomotion intelligence, task intelligence, and interaction intelligence, along with workshops and sub-forums. The report also reviewed milestones such as the Elf G2 and Lingxi X2 deployments in production lines, stores, and hotels, and the 20000th general-purpose robot rolling off the line.

紫牛新闻(扬子晚报网) · 约 5 分钟

2026-10-21 · 本站更新

大模型与软件Industry Updates

StepFun Open-Sources Step 5 Preview: 600B-Parameter MoE with 1M Context, Built for Long-Horizon Agents

On 15 10, 2026, StepFun open-sourced the Step 5 Preview model weights, available in both FP8 and BF16 formats. The model uses a sparse MoE architecture with 600B total parameters and 27B activated parameters, contains 92 Transformer blocks, and supports up to 1M tokens of context; the company says it targets long-horizon Agent tasks in software engineering and professional knowledge work. Commercial use requires reviewing the additional conditions of the StepFun Model License 1.0.

七牛云 · 约 10 分钟

大模型与软件Industry Updates

MiniMax Releases M3.1 Flash: Focused on Agentic Task Performance and Cost

MiniMax officially released MiniMax M3.1 Flash. Officially, it previously ran under the anonymous model name Space Bunny, cumulatively processing over 200 trillion tokens. The new model focuses on agentic tasks, coding, Computer Use, and cybersecurity, with officially published benchmark results including AutomationBench-AA, BankerToolBench 76.2, DeepSWE v1.1 73.7%, OSWorld V2 61.7, ALE 29.6, and CyberGym 82%. Officials claim nearly 3x faster online inference speed and a 70% cost reduction versus the previous generation, priced at input $0.15, output $0.60, cached input $0.015 (per 1M tokens). The model is now available via MiniMax API, M Plan, and MiniMax Code, with weights to follow.

同一事件另有 1 篇报道

MiniMax Research · 约 5 分钟

具身智能机器人Industry Updates

Zhiyuan Expedition A3 Ultra Lands at Wuhan Riverside: Full-Size Humanoid Robot Smart Retail Warehouse Begins Real-World Operations

Zhiyuan, partnering with Wuhan Culture & Tourism Group and Zhongyun Weixun, has built a full-size bipedal humanoid robot smart retail warehouse, with Expedition A3 Ultra conducting real-world commercial operations at the "Jiangyu" Bridge-Pier Movie Camp on Hankou Riverside. As of 10 19, it has run stably for 18 days, completed a cumulative 827 orders, with an overall grasping success rate above 97%, and during visitor peaks achieved over one hundred single-day orders with a grasping success rate of 100%.

同一事件另有 1 篇报道

证券时报·e公司(转载人民财讯) · 约 4 分钟

具身智能机器人Industry Updates

2026 Global Developer Pioneer Conference Opens in Shanghai on 10 23, Focusing on Real-World Embodied Intelligence Deployment

According to China City Net, the 2026 Global Developer Pioneer Conference and International Embodied Intelligence Skills Competition will be held from 10 23 to 25 at the Zhangjiang Science Hall in Pudong, Shanghai. Centered on "real machines, real scenarios, real-world training," the conference has announced ten highlights: exhibitors include complete machine makers such as Zhiyuan (Agibot), Unitree, and Fourier; the competition features 9 tracks and 23 events; over 100 open source communities will participate; and a vertical large model evaluation benchmark and China Mobile's Spatio-Temporal Intelligence model series will be released.

中国城市网 · 约 7 分钟

大模型与软件Industry Updates

阶跃星辰开放 Step 5 Preview 权重,600B 参数稀疏 MoE 模型上线 Hugging Face 与 ModelScope

阶跃星辰于 2026 年 10 月 15 日正式开放 Step 5 Preview 模型权重,含 BF16 与 FP8 两个版本,并附技术博客、开源开发工具及部署指南,权重已上线 Hugging Face 和 ModelScope。该模型面向长程 Agent 任务设计,采用稀疏 MoE 架构,总参数 600B,每 token 激活 27B,支持 100 万 token 上下文。据 OpenRouter 截至 10 月 15 日数据,其 Token 用量周榜及日榜均列全球第二。

科创板日报 · 约 2 分钟

2026-10-20 · 本站更新

大模型与软件Industry Updates

Meta 开放 Muse Spark Open:816B 参数 MoE 模型的 Hugging Face 申请与 SGLang/vLLM 部署路径

Meta 于 2026 年 10 月 15 日发布开放权重模型 Muse Spark Open,总参数 816B、每 token 激活 42B,上下文窗口 1,048,576 token,原生支持文本、图片、音频、视频输入。本文依据官方部署文档整理权重体积(BF16 约 1.65TB、NVFP4 约 537GB)、B300 GPU 与 Apple MLX 参考硬件、SGLang/vLLM 启动命令、reasoning_effort 参数及 muse-model-license 的 MaaS 营收条款。

CSDN博客(七牛云行业应用) · 约 4 分钟

具身智能机器人Industry Updates

Unitree Robotics Stock Jumps Over 17% in a Single Day, Market Cap Returns to 2000 Billion Yuan, First Ecosystem Conference Set for 10 22

Unitree Robotics-W (688836.SH) surged intraday on 10 19. As of 14:33, the share price stood at 499.99 yuan per share, up 17.10%, with total market cap rebounding to 2022.28 billion yuan — the largest single-day gain since its STAR Market debut on 8 19, excluding the IPO day. Multiple humanoid robot concept stocks hit the daily limit up that day. The company's first ecosystem conference is scheduled for 10 22 in Hangzhou, with a preliminary Q3 report disclosure date of 10 29.

同一事件另有 1 篇报道

新浪财经(转载21世纪经济报道) · 约 9 分钟

大模型与软件Industry Updates

Qianwen Office E-commerce Edition Launches, AI Agent Boosts Merchant Operations Efficiency

On 10 14, 2026, Qianwen Office CEO Chen Yusen disclosed at the e-commerce edition launch event that more than two months after launch, the product's paid user share in the e-commerce sector exceeded 20%. A merchant case presented at the event showed that after adopting the AI agent, new store preparation time was compressed from one month to 7 days, initial investment dropped from over 1 thousand yuan to a few hundred yuan in Token costs, and daily operations work takes only 2 to 3 hours.

软盟资讯 · 约 5 分钟

2026-10-16 · 本站更新

Industry InsightsIndustry Updates

Meta releases Muse Wonder model series, focusing on agent tasks and professional-domain capabilities

According to Meta's official press release, Meta launched two models, Muse Wonder and Muse Wonder Full, on 10/13, describing them as the company's most capable models to date, achieving industry-leading performance on personal and professional agent tasks. The new models strengthen professional-domain knowledge in law, finance, and healthcare as well as cutting-edge coding abilities, and introduce two new features, Active Memory and Active Focus. They are available starting today via Meta Model API and Muse Code, with subscribers getting first access.

腾讯新闻 / PANews · 约 3 分钟

Compute and ServersIndustry Updates

NVIDIA Opens Vera Rubin Rack Designs at OCP Summit and Releases Data Center Specs with Multiple Partners

At the OCP Global Summit in San Jose, NVIDIA announced it will open the key rack and compute tray designs of the Vera Rubin platform to OCP via the MGX 1.2 architecture, covering blind-mate liquid cooling, a 5000-amp busbar, and 250-kilowatt rack capability. It also released the Vera CPU rack specification, the Open Data Center Specification v0.7, and a Power Grid Resilience Specification, and jointly introduced the USFP optical module specification with Cisco. This article outlines the key specifications and ecosystem developments for data center planners.

同一事件另有 1 篇报道

报价宝 · 约 10 分钟

Industry InsightsIndustry Updates

Microsoft Announces Scaling Azure Maia Deployments to Multi-Gigawatt Level Starting 2027, Maia 300 Targeting Primary Inference Workloads

Microsoft announced that starting in 2027, it will scale its Azure Maia series AI chip system deployments to the multi-gigawatt level, positioning it as the primary inference system for Microsoft's products and services. According to a Sina Finance flash report, Maia 200 is already in production environments serving workloads such as Microsoft Copilot; Maia 300 is scheduled to begin deployment in 2027, targeting long-context inference, low-latency conversation, and multi-step agent tasks. Microsoft expects its per-megawatt token processing throughput to improve by 40% compared to leading GPU systems.

新浪财经 · 约 6 分钟

Product Selection Assistant

让需求变成可修改的筛选条件

依据已录入参数生成条件,结果可查看和修改。帧率与性能需核对实测,不能仅凭算力数值保证。