晨报 · 2026-07-19

9点1氪丨ofo停更5年突然发文，运营主体仍处存续状态；苹果市值重返全球第一；乐事回应“蓝色薯片”来源

今日热点导览 月之暗面正式推出开源模型Kimi K3 智谱ARR达到10亿美元，半年增长15倍 国际足联：世界杯冠军将有戒指，每枚均有独立编号 国内成品油价7月17日晚上调 A股存储板块龙头股遭遇两个跌停 TOP3大新闻 ofo停更5年突然发文，运营主体仍处存续状态 7月15日，沉寂近5年的ofo小黄车官方微信公众号突然更新。此次公众号发布的内容，是一则皮肤病患者援助招募信息。虽然与共享单车业务并无关联，这次更新仍然迅速引发关注。不少网友提及尚未退回的押金问题：“我的199元什么时候退？”“退款队伍还排着吗？”“这么多年...

印奇在 WAIC 2026 开幕式主论坛发表主题演讲：当智能体走进物理世界

2026 世界人工智能大会（WAIC 2026）于 7 月 17 日正式开幕。作为全球人工智能领域的顶级盛会，本届大会以“智能伙伴 共创未来”为主题。阶跃星辰董事长、千里科技董事长印奇作为特邀嘉宾出席大会开幕式并在大会主论坛（上午场）发表主题演讲《当智能体进入物理世界》。回顾 15 年 AI 创业历程，他表示，AI 创业已从小众赛道成为全球重要共识。今天的 AI 研究者和创业者既站在几代科学家持续探索的肩膀上，也受益于信息技术、移动互联网、云计算、半导体、新能源等产业积累，能够参与这场时代变革无比幸运。 印奇判断，2026 年模型能力正在跨越关键临界点，AI 已从只能连续执行数秒...

氪星晚报｜阿里1688将推出AI时代B2B交易互联互通开放标准；英特尔与Google Cloud宣布深化战略合作；铁路部门试点提前60天预约购票服务

大公司： 成大生物：流感病毒裂解疫苗（高剂量）进入I期临床试验 36氪获悉，成大生物公告，公司全资子公司成大生物(本溪)研发的流感病毒裂解疫苗（高剂量）已获国家药监局临床试验批准，并完成I期临床试验筹备工作，正式进入I期临床试验阶段。该疫苗抗原含量为常规剂量四倍，适用于60岁以上老年人群及高风险人群，国内尚无同类产品获批上市，有望填补市场空白。 苏宁易购携手人工智能合作伙伴，布局家用服务机器人 36氪获悉，近日，苏宁易购旗下碧英科技入围工信部人工智能揭榜挂帅项目，将联合产业生态合作伙伴研发智能家庭陪护机器人。该项目面向老龄化背景下的居家照...

Index Ventures 联合创始人 Neil Rimer 认为 AI 财富将面临"再分配"

Index Ventures 联合创始人 Neil Rimer 表示，围绕 AI 积累的巨额财富将面临"某种形式的再分配"，无论是自愿还是强制。他呼吁科技领袖在推动自愿再分配中发挥主导作用。与此同时，美国慈善捐赠总额虽创新高，但捐赠人数持续下降，而加州正考虑对亿万富翁征收 5% 的一次性财富税。（AIHOT源：TechCrunch：AI（RSS））

Apple 起诉 OpenAI：诉讼背后是竞争焦虑还是时机博弈？

Apple 对 OpenAI 提起诉讼，指控其存在多项不当行为，尽管许多专家认为部分指控属于行业惯例。此举正值 Apple 发布新版软件公测版（以新 Siri AI 为核心）之际，外界猜测 Apple 究竟是担忧 OpenAI 成为潜在竞争对手，还是想利用 OpenAI 的弱势期获利。（AIHOT源：The Verge：AI（RSS））

月之暗面在GTC 2026披露Kimi K2.5技术路线：用MuonClip、线性注意力与Agent Swarm重构三大基础组件

月之暗面CEO杨植麟在GTC 2026演讲中提出用MuonClip优化器替代Adam，可将数据利用效率提升近一倍。同时推出Kimi Linear线性注意力，在百万Token上下文下全面超越全注意力；Agent Swarm已支持300个Agent并行工作。（AIHOT源：X：宝玉 (@dotey)）

八天四款前沿模型发布，Kimi K3 跻身第三

过去八天内，Grok 4.5、GPT-5.6、Muse Spark 1.1 与 Kimi K3 四款前沿模型相继发布，使 Artificial Analysis Intelligence Index 得分超 50 的实验室从 6 月初的 2 家增至 6 家。（AIHOT源：X：Artificial Analysis (@ArtificialAnlys)）

美团LongCat发布LoHoSearch：更难搜索智能体基准

美团LongCat推出LoHoSearch，一个基于762万实体维基百科知识图谱自动生成问题的搜索智能体基准，旨在解决BrowseComp等现有基准趋于饱和的问题。在11个前沿模型测试中，最佳得分仅34.74%，远低于当前模型在BrowseComp上约90%的成绩；上下文策略仅带来+6.8个百分点的提升。该基准包含544道问题、11个领域，采用树与图结构，已开源。（AIHOT源：X：美团 LongCat (@Meituan_LongCat)）

LLM cliché highlighter：一款识别AI写作套话的检测工具

Simon Willison 用 Fable 5 开发了一款 LLM cliché highlighter 应用，用于高亮 LLM 生成文本中常见的十种套话模式，例如"no fluff， no filler， no jargon"这类陈词滥调。该工具旨在帮助读者快速识别并过滤掉充斥在文章中的 AI 写作风格化表达。（AIHOT源：Simon Willison 博客）

苹果与 OpenAI 法律战升级：约 40 名前员工收到苹果律师函

苹果已向约40名就职于OpenAI的前员工发出律师函，要求保存相关文件。此前苹果起诉OpenAI及两名前员工，指控其通过挖角获取商业机密以加速AI硬件研发。苹果称已有超400名前员工在OpenAI工作，正寻求法院禁令阻止OpenAI使用苹果信息并要求归还机密。（AIHOT源：IT之家（RSS））

OpenAI 提出 AI 时代记分卡："有用智能每美元"衡量实际工作价值

OpenAI 提出"Useful Intelligence per Dollar"（有用智能每美元）作为衡量 AI 投资回报的核心指标，从完成的有用工作量、成功任务的实际成本、结果可靠性三个维度评估。（AIHOT源：OpenAI：官网动态（RSS · 排除企业/客户案例））

首届"小有可为"大赛乡村教育一等奖作品"智绘科普"技术拆解

首届"小有可为"大赛乡村教育赛道一等奖作品"智绘科普"采用 Qwen3.5-397B-A17B 大语言模型与 Manim 动画引擎，通过多Agent分阶段协作与自动修复机制，将知识主题转化为可控、可编辑的教学动画。系统包含规划、草稿、实现、审查、合成五个阶段，渲染失败时可自动提取日志并修复，该工程范式可迁移至其他赛道。（AIHOT源：公众号：通义实验室（千问））

Schema Harness 在 ARC-AGI-3 公开集上取得约 99% 成绩

Schema 框架在 ARC-AGI-3 公开集上，使用 Claude Opus 4.8 和 Fable 5 达到 99% RHAE 分数，使用 GPT-5.6 Sol 达到 95.35%。该框架不修改模型权重，而是将原始观测转化为可编辑程序，联合解决状态归因和机制发现问题。此前最强模型 GPT-5.6 Sol 在半私有集上仅得 7.78%。（AIHOT源：Hacker News 热门（buzzing.cc 中文翻译））

NVIDIA 发布 Nemotron 3 Embed 系列，8B 版本在 RTEB 基准上排名第一

NVIDIA 发布 Nemotron 3 Embed 系列，包含三个开源 checkpoint，其中 8B-BF16 版本在 RTEB 基准上以 78.46 的平均 NDCG@10 排名第一。1B-NVFP4 版本在 Blackwell 上吞吐量比 BF16 高 2 倍，精度保留 99.5%，所有模型最大序列长度 32，768 tokens。（AIHOT源：MarkTechPost（RSS））

通义实验室发布 Wan-Streamer v0.2，端到端响应延迟仅 550ms

通义实验室发布 Wan-Streamer v0.2，这是一款将"听、看、说、演"统一进单个 Transformer 的端到端全模态模型。其端到端响应延迟仅 550ms，输出分辨率从 v0.1 的 192×336 提升至 640×368 @ 25FPS，并采用 Thinker-Performer 双通路架构在提升画质的同时维持了极低延迟。（AIHOT源：公众号：通义实验室（千问））

How We Beat HotSpot Performance (By Cheating, But Not Like That)

Stratagems #17: Alex Set an AI Bait. The Catch Wasn't Code — It Was Someone Who Shouldn't Have Been Watching.

FLUX on a 4070 Graphics Card 🖼️

Every AI-built site looks the same, so I built a skill that locks taste before any code is written

Why I built a SQL client when 10 already exist — then let AI agents into it

Anthropic Engineering: An update on recent Claude Code quality reports

Over the past month, we’ve been looking into reports that Claude’s responses have worsened for some users. We’ve traced these reports to three separate changes that affected Claude Code, the Claude Agent SDK, and Claude Cowork. The API was not impacted. All three issues have now been resolved as of April 20 (v2.1.116). In this post, we explain what we found, what we fixed, and what we’ll do differently to ensure similar issues are much less likely to happen again. We take reports about degradation very seriously. We never intentionally degrade our models, and we were able to immediately confirm that our API and inference layer were unaffected. After investigation, we identified three different issues: On March 4, we changed Claude Code's default reasoning effort from high to medium to reduce the very long latency—enough to make the UI appear frozen—some users were seeing in high mode. Th...

Anthropic Engineering: Scaling Managed Agents: Decoupling the brain from the hands

Get started with Claude Managed Agents by following our docs . A running topic on the Engineering Blog is how to build effective agents and design harnesses for long-running work . A common thread across this work is that harnesses encode assumptions about what Claude can’t do on its own. However, those assumptions need to be frequently questioned because they can go stale as models improve. As just one example, in prior work we found that Claude Sonnet 4.5 would wrap up tasks prematurely as it sensed its context limit approaching—a behavior sometimes called “context anxiety.” We addressed this by adding context resets to the harness. But when we used the same harness on Claude Opus 4.5, we found that the behavior was gone. The resets had become dead weight. We expect harnesses to continue evolving. So we built Managed Agents: a hosted service in the Claude Platform that runs long-horizo...

Claude Blog: New in Claude Managed Agents: self-hosted sandboxes and MCP tunnels

Starting today, Claude Managed Agents can operate in a sandbox you control and connect to your private Model Context Protocol (MCP) servers. Both the sandbox where an agent executes tools and the services it reaches run within the established boundaries of your enterprise, under your security and runtime controls. The sandbox runs on your own infrastructure, or with managed providers like Cloudflare , Daytona , Modal , or Vercel to handle the compute and isolation for you. On the Claude Platform, self-hosted sandboxes is available in public beta and MCP tunnels in research preview ( request access ). Self-hosted sandboxes: keep agent execution within your perimeter A self-hosted sandbox lets a Claude Managed Agent execute tools on infrastructure you control or with a managed sandbox provider. Code execution, sensitive files, packages, services, and data stay within your enterprise perime...

Claude: Beginning July 20, Claude Fable 5 will be included in all Max and Team Premiu...

Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.｜@claudeai · 热度 38674 · 2026-07-18

Thibault Sottiaux: GPT-5.6 Sol confirmed to be an extremely good model https://t.co/nZaAnFhsPX

GPT-5.6 Sol confirmed to be an extremely good model https://t.co/nZaAnFhsPX｜@thsottiaux · 热度 12505 · 2026-07-18

Thibault Sottiaux: Oops..

Oops... I did it again. Enjoy reset usage limits for all paid users for Codex and ChatGPT Work. Super grateful for an incredible team who is iterating at lightspeed and keeping the infra up as we scale faster than ever. Enjoy the weekend! https://t.co/WIuS07doKA https://t.co/3NwNqHG7TV｜@thsottiaux · 热度 12087 · 2026-07-18

Swyx: btw if you havent set your {codex | claude | gemini | devin} automations to a...

btw if you havent set your {codex | claude | gemini | devin} automations to autoresearch how to improve your seo/aeo every week you are really truly missing out on free, should-be-commoditizing-but-weirdly-untapped alpha｜@swyx · 热度 1118 · 2026-07-17

Thariq: building prototypes of mockups, schemas, data models, proof of concepts, etc

building prototypes of mockups, schemas, data models, proof of concepts, etc. is the best way to avoid spending tons of tokens before realizing you don't want the output https://t.co/BjYeXC7zfk｜@trq212 · 热度 757 · 2026-07-17

Aaron Levie: This post is key

This post is key. The cheaper AI gets, the more opportunity there is for the entire ecosystem - especially including end-customers - to benefit. Everything is bottlenecked by being able to successfully and cost effectively deploy AI in real workloads. Any time we can lower the cost of AI, the total usage goes up. When that happens, the value accrues to all layers of the stack, as noted. The only modification is I think as AI gets more efficient, demand even for the frontier closed models can go up too. You often need the strongest model possible for the orchestration of a task, then you can farm out work to cheaper or more tuned models for the bulk of tokens. This efficiency gain can ironically lead to even more frontier spend in the process because you can cost effectively deploy AI against more tasks. The thing at most risk is margins, which is fair, and in general it seems that intelligence eventually should converge with the margins of the infrastructure stack. Anyway, super fun times to see competition play out to drive down the cost of AI.｜@levie · 热度 448 · 2026-07-17

Peter Steinberger: It's both amazing and painful to watch codex use browser + computer use to op...

It's both amazing and painful to watch codex use browser + computer use to open Chrome, go to my PR, tap on comment and wrangle with the macOS picker - all TO UPLOAD AN IMAGE. GitHub has no API doesn't stop anyone. I let my codex run in VMs so they don't steal app focus. https://t.co/Dd5KzNs6E5｜@steipete · 热度 383 · 2026-07-18

Peter Steinberger: ya'all made me go crazy with codexbar icon customization issues, so I built a...

ya'all made me go crazy with codexbar icon customization issues, so I built an editor. (by me, I mean codex) https://t.co/edAthj1eZw｜@steipete · 热度 343 · 2026-07-17

Madhu Guru: The reason enterprises struggle to go beyond basic chat bots is the talent ga...

The reason enterprises struggle to go beyond basic chat bots is the talent gap to build harnesses and evals. 1. Evals: do you clearly understand your use cases and can you replicate that in the form of offline and online evals. Do the evals express your ambition and do they push the jagged frontier of the models? Do they help you pick the right models on the quality-cost-latency curve? 2. Harness: do you have a system that manages routing, multiagency orchestration, context management, tool calling, memory that is independent of the models? 3. Talent: do you have the talent to build this all out on the frontier? This is the scarcest piece.｜@realmadhuguru · 热度 319 · 2026-07-17

mattpocock/skills

Skills for Real Engineers. Straight from my .agents directory.｜Shell · ⭐176519 · Search 增补 · 近快照 +911星 · 创建 2026-02-03

Graphify-Labs/graphify

AI coding assistant skill (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more). Turn any folder of code, SQL schemas, R scripts, shell scripts, docs, papers, images, or videos into a queryable knowledge graph. App code + database schema + infrastructure in one graph.｜Python · ⭐90909 · Search 增补 · 近快照 +723星 · 创建 2026-04-03

DietrichGebert/ponytail

Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.｜JavaScript · ⭐85569 · Search 增补 · 近快照 +373星 · 创建 2026-06-12

VoltAgent/awesome-design-md

A collection of DESIGN.md files analysis by popular brand design systems. Drop one into your project and let coding agents generate a matching UI.｜Unknown · ⭐102929 · Search 增补 · 近快照 +14773星 · 创建 2026-03-31

Leonxlnx/taste-skill

Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop｜JavaScript · ⭐64924 · Search 增补 · 近快照 +331星 · 创建 2026-02-19

LG monitors silently install software through Windows Update without consent

Regressive JPEGs

GPT-5.6 used a prompt to close a 30-year gap in convex optimization

What AI did to stackoverflow in a graph

The Zilog Z80 has turned 50

不能输的理由：DK 2-1 击败卫冕冠军 Gen.G 晋级 2026 EWC《英雄联盟》总决赛，KC 战胜 T1 捍卫欧洲主场荣耀

IT之家 7 月 19 日消息，在 7 月 18 日举行的 2026 EWC 电竞世俱杯《英雄联盟》项目半决赛中，LCK 赛区四支队伍中“赛前最不被人看好的”DK 战队以 2-1 的成绩战胜了 EWC 卫冕冠军 Gen.G ，晋级最终决赛，锁定至少 34 万美元（IT之家注：现汇率约合 230.7 万元人民币）奖金。 至此，赛前被人寄予厚望的 BLG、HLE、AL、T1、 Gen.G 、G2、LYON 已全部遭遇爆冷。 这是 ShowMaker 率队时隔 1723 天重返国际赛决赛舞台，重铸 DK 荣光。上一次 DK 晋级决赛还是 2021 全球总决赛（DK 以 3:2 击败 T1...

英伟达 N1X 首个 Cinebench 2026 跑分曝光：微软 Surface Ultra 工程样机现身，单核成绩接近苹果 M3 Max

IT之家 7 月 18 日消息，爆料者 @ExoticSpice101 今日放出了英伟达 N1X 芯片首个 Cinebench 2026 测试数据。 测试平台是一台 Surface Ultra 工程样机，该机在测试中获得单核 540 分、多核 5771 分的成绩，单核性能据称“与苹果 M3 Max 相当”。当然，英伟达和微软尚未公布 N1X 的官方性能数据，该数据仅做参考。 Cinebench 2026 是 Maxon 面向新一代处理器推出的基准测试工具，主要用于评估 CPU 在渲染等多线程计算任务中的性能表现。由于工程样机可能受到固件、驱动、电源策略以及散热设计影响，最终产品...

英伟达 RTX Spark 首批 Win11 Arm 驱动发布：最高 6144 个 Blackwell 核心，N1X PC 预计今秋登场

IT之家 7 月 18 日消息，英伟达已经发布了专为 RTX Spark PC 准备的 Windows 11 驱动程序，这意味着距离搭载 N1X 芯片的设备正式上市又近了一步。 目前该驱动主要面向开发测试设备，并非普通消费者产品，但其中包含的硬件信息进一步确认了 RTX Spark 平台部分 GPU 配置。 英伟达官网出现了一款版本号为 616.00 的 Windows 11 Arm64 原生驱动程序。对安装文件进行分析后发现，其中包含名为“ nv_surface_woa.inf ”的配置文件。Windows 驱动中的 INF 文件通常用于记录支持设备的硬件信息、设备 ID 以...

火速上架！空降即登顶Arena，Kimi K3当日登陆「模型广场」

点击查看原文>

中国 AI 产业核心要素出海发展白皮书

点击查看原文>

为什么 AI Agent 拿到数据却不会推理？可观测对象图语义层的设计与开源实践｜AICon深圳

点击查看原文>

Reviewing AI Code Is Not A Viable Argument

Regressive JPEGs

GitRoot

Lobsters Interview with matheusmoreira about Lone Lisp

"Half a Second" - a book on the XZ backdoor

The Download: perimenopause misinformation and China’s latest AI leap

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. There’s a lot of hype around perimenopause. Don’t buy it. Perimenopause used to be considered taboo, but not anymore. Thanks at least in part to TV doctors and...

There’s a lot of hype around perimenopause. Don’t buy it.

Perimenopause has entered the chat. Perimenopause—and its better-known relative, menopause—used to be considered taboo. Not anymore, thanks at least in part to TV doctors and social media influencers. Perhaps it’s my age, but these days, both my algorithm and my conversations with friends increas...

The risk of weather data sabotage is rising

Every morning, airline dispatchers, grid operators, and farmers around the world make decisions based on the same thing: a weather forecast. While these forecasts are something that most people glance at for two seconds, weather predictions influence major strategic decisions in many industries, ...

Waymo says San Francisco service has resumed after one-hour pause

This isn’t the first time power outages have caused issues for Waymo.

Kimi: Threat or menace?

Chinese company Moonshot AI released a new version of its Kimi model this week, prompting concern about "full AI communism."

All the EVs that were discontinued or killed off in the U.S. this year

Th Honda Prologue will no longer be sold in the U.S., joining a growing list of EV models to exit the market this year.

Dave Eggers told OpenAI staff that ChatGPT was ‘silencing an entire generation’

Last year, Sam Altman invited author Dave Eggers to give a talk to around 200 OpenAI staffers. The man has written countless novels, screenplays, pieces of journalism, started McSweeney's, and founded multiple schools and nonprofits that support writers and the arts more broadly. So one might exp...

Google might not kneecap the Pixel 11a with an old processor

Mystic Leaks suggests that the Pixel 11a will return to featuring a flagship-grade processor with the Tensor G6. Rather than the Tensor G5 found in the Pixel 10 and 10 Pro, the Pixel 10a shipped with the previous generation Tensor G4. That was a huge disappointment since, typically, the Pixel a l...

Google is open-sourcing its 3D emoji

Now, if you want to, you can use Google's 3D emoji in your own creations. The company shared some details about how it went about designing the little pictograms and why, as part of World Emoji Day on Friday. Things you might not necessarily worry about in a 2D illustration suddenly become very i...

小米 NAS 拆解与初步上手：它会是你的第一台 NAS 吗？

小米智能存储最适合两类用户：家里已经有较多米家设备的人，以及小米手机用户。 查看全文

本周看什么 | 最近值得一看的 10 部作品

📅本周新预告《蓦然回首》正式预告7月15日，电影《蓦然回首》发布了正式预告，将于9月11日在日本上映。是枝裕和执导、编剧、剪辑，出口夏希、莳田彩珠主演，菅田将晖、宫藤官九郎、野吕佳代、山田真步等出演， ... 查看全文

具透 | watchOS 27 公测版一览：动态应用网格、Liquid Glass 微调……

AI 之外，watchOS 27 中还有这些新功能。 查看全文

Doubao-Seed-Evolving升级：1M上下文来了！

火山引擎推出动态升级模型 Doubao-Seed-Evolving，百万上下文、长任务性能更强，统一 ID 自动免迁移升级，适配代码、多轮对话，平台限时调用八折、9.9 元起套餐。

ThreadLocal 不香了？ScopedValue才是王道？

ThreadLocal 基本概念 应用案例 StructuredTaskScope ScopedValue 基本概念 基本用法 源码分析 小结 ThreadLocal是一种实现将变量在各线程之间隔离的

技术管理笔记：让我“燃尽”的一天

大家好，我是拭心。 今天是周六，睡了一个好觉，终于可以坐下来深度思考一下了。 最近事情非常多，每天都在项目、需求、技术、质量和团队之间来回切换。 身处其中的时候，更多是在解决一个又一个具体问题

美对伊展开新空袭，打击海峡航运威胁武装

美国于周六对伊朗发动新一轮空袭。 美国中央司令部称，此次打击目标包括两类设施：一是可用于威胁霍尔木兹海峡商船航运的相关装备...

美中央司令部：美军开始对伊朗发动新一轮空袭

当地时间7月18日，美国中央司令部表示，美国东部时间当日下午6时，美军奉三军统帅之命，开始对伊朗发动新一轮空袭。 美军称，此次行动旨在进一步削弱伊朗威胁霍尔木兹海峡...

特朗普：痛心美军士兵遇难，绝不允许伊朗拥核，无视协议停摆

美国总统特朗普称两名美军人员在约旦遇难一事“令人万分痛心”。 他表示：“我们不愿看到这样的悲剧发生，这些军人是为国效力。”同时再次强调，伊朗“不能...

伊朗革命卫队称打击三国美军目标

当地时间7月18日，伊朗伊斯兰革命卫队发布了近期打击美军基地的画面。画面内容显示，在“纳斯尔-2”第17至20波行动中，使用“海巴尔·谢坎”“佐勒菲卡尔”“征服者-110”“哈吉·卡西...

仍有近200万美国人至少半年未能进入劳动力市场

6月，失业半年及以上的美国求职者占比达到27.3%，逼近2021年末以来的最高值。多项核心指标显示，美国劳动力市场整体表现向好：经济已连续四个月新增就业岗位...

WAIC看了一圈，这家公司的机器人在认真打工

机器人真正具备了干活的完整能力

WAIC 2026现场直击：万兴科技旗下万兴剧厂“无限画布”、新品Filmora.AI与Virbo AI火爆首秀

滴滴在WAIC分享：AI升级出行体验 好服务连接消费烟火气