数据中心监管压力加剧！美国多州州长候选人支持暂停建设
Cancel your $99.99 Microsoft 365 subscription and get the same apps for life for $49.97
TL;DR: A Microsoft Office 2024 Pro Lifetime License gives you six apps with no monthly fees, and it’s only $49.97 (reg. $249.99).
A Microsoft 365 Personal subscription costs $99.99 a year, every year for apps you’ll likely be using forever. Microsoft Office 2024 Professional Plus is a one-time purchase that gives you many of the same apps, and a lifetime license is only $49.97 (reg. $249.99).
This license gives you access to Word, Excel, PowerPoint, Outlook, OneNote, and Access all installed directly on your PC, and none of them depend on the cloud to work.
Office 2024 has low system requirements and runs on a wide range of Windows 10 and Windows 11 PCs. You also get dark mode across the apps, along with tools for analyzing data in Excel and managing databases in Access. Microsoft sends regular security updates for as long as the product is supported.
More than 10,000 people have bought this license, and buyers rate it 4.8 out of 5 across 380 reviews.
Get Microsoft Office 2024 Pro Plus on sale for $49.97.
Offer ends Nov. 1 at 11:59 p.m. PT.
Microsoft Office 2024 Professional Plus for Windows: Lifetime LicenseSee Deal
StackSocial prices subject to change.
184 Blog Posts To Learn About Megafansesports
Let's learn about Megafansesports via these 184 free blog posts. They are ordered by HackerNoon reader engagement data. Visit the Learn Repo or LearnRepo.com to find the most read blog posts about any technology.
MegaFans Esports is a platform likely focused on mobile esports, fostering competitive gaming and community engagement. It matters by democratizing esports participation and providing opportunities for players to compete and earn in a growing mobile gaming landscape.
1. The Future of Gaming Writing Contest by Megafans x HackerNoon
The Future of Gaming Writing Contest runs from April 1 to June 30th w/$5300 in cash prizes for the top 5 stories submitted each month.
2. What is EOMM, Engagement Optimized Matchmaking?
Unlike SBMM, EOMM emphasizes engagement rather than skill. How does this match-making algorithm work?
3. More Than Just The Future Of Gaming
Gaming has provided an opportunity for people to learn about logistics, economics and logistics by their very existence, making the future of gaming interesting
4. The Dark Side of Roblox: Child Exploitation and Sex Games
Roblox is a very scandalous platform, for example, you can find sex games (condo games) here. Also, child labor exploitation is present.
5. 8 Best Cozy Games to Play When the World is Falling Apart
The idea behind cozy games is the combination of appealing colors and atmospheric music. This article lists some of the best cozy games of recent years.
6. ProVolver Review: VR Haptic Feedback at its Finest
ProTubeVR is one of my favorite companies in the VR peripheral market, and their newest device, the ProVolver, could be their best one yet.
7. The Best Weapon Skins in PUBG
PUBG Mobile is one of the best mobile games in the world with a flourishing competitive scene. Excluding its high-level gaming experience and stunning graphics, there is one more thing that made this game more special and famous & that is its end-less cosmetics.PUBG is an example of high creativity plus a symbol of endless ideas and schemes. All the weapon skins and outfits of this game are customized decently. These skins represent the research of a collection of different themes and events according to the latest seasons.There are several skins of all the weapons available in the game. Not all but some of them can level up to their final form that we are discussing in detail.
8. Is Cloud Gaming Doomed Because of Physics?
Cloud gaming is a hot topic today but gamers speak negatively about input latency. I discuss the journey of input latency from cloud servers to users' monitors.
9. The Fallout Games in Chronological Order
For anyone looking to catch up on the franchise’s lore, it would be useful to learn about the Fallout games in Chronological Order.
10. How to Build an Arduino Starship Game Controlled by Joystick and Computer
In this article, we will develop an Arduino Starship game which will be displayed on an LCD display 16x2.
11. How I Let an AI Code a Game For Me!
What if I would ask an artificial intelligence to writa game for me? Is it hard? Will it work? I'm using Github Copilot to try it out.
12. 5 Best Final Fantasy Games of All Time Ranked by Sales
Final Fantasy VII, Final Fantasy XV, Final Fantasy XIV, Final Fantasy X, and Final Fantasy VIII are the 5 best-selling games in the Final Fantasy franchise.
13. 5 Elden Ring PvP Builds that Dominate the Battlefield
Elden Ring's PvP scene is already taking off. Here are 5 great PvP builds for duelists and Bloody Fingers in The Lands Between.
14. How I Developed the Classic Pong Game on an Arduino Board
I have figured out a way to develop the popular pong game on the Arduino board.
15. 10 Best Steam Games of All Time Ranked by Sales
Steam is one of the largest gaming pla
Amid AI Bans, Microsoft and OpenAI Plan a 'Joyful' Hour of AI for Kids
Read more of this story at Slashdot.
Which Tech Companies Started Using Trump's Phrase 'Super Intelligence'?
Read more of this story at Slashdot.
SystemD's Creator Opens Signups for Private Preview of Amutable's Cryptographically Verified Linux Systems
Read more of this story at Slashdot.
A社/O社/谷歌/微软联合起草拒绝AI训练声明 网站只需要添加声明即可阻止内容用于AI训练
天数智芯天垓300实测对标H200，国产GPU接下来拼什么？
在实测场景中达到H200的性能水平后，国产GPU的竞争，开始进一步转向系统效率、软件适配与实际交付。
本周的天数智芯媒体开放日上，最新一代天垓300与H200的实机对比，是最直观的性能展示。让GPU与专用芯片组成异构系统，是对推理效率的新探索。软件栈引入Agent与Skills，则试图解决客户部署、适配和优化的工程难题。
天数智芯表示，今年完成的一组第三方测试中，在典型PD分离部署下，天垓300折算单卡端到端吞吐达到H200的1.05倍。这表明国产通用GPU在特定高端推理负载中，已经拥有与H200直接竞争的性能基础。
但从性能接近到客户采购，再到持续扩容，中间仍隔着系统效率、软件适配和稳定交付。当硬件具备竞争力，国产GPU怎样把这种能力转化为客户持续使用的理由？当推理催生更多专用芯片，通用GPU的独特价值是什么？
从云端天垓300到端侧彤央系列，天数智芯也在拓展产品布局，将竞争延伸到不同形态的算力需求中。
天垓300对比H200实测，性能小幅领先
过去很长一段时间，峰值算力是国产AI芯片受到关注的重要指标。但对最终用户而言，设备每秒真正产出多少有效Token，比参数表上的理论算力更接近采购决策。
天数智芯披露的第三方测试覆盖四大主流模型、三类场景。其中，1.05倍指典型PD分离部署下，折算单卡的整体端到端吞吐，反映的是特定模型与部署条件下的系统表现。
另外，根据天数智芯性能实验室的数据，在GLM-5.3的64K长上下文场景下，保障同等用户体验时，天垓300的MoE吞吐达到H200的1.15倍。在8K至128K长上下文的Agent场景中，首字延迟较H200降低约10%至30%。
前者关系到系统能够承接多少请求，后者关系到用户发出请求后需要等待多久，它们对应不同工作负载与性能指标。
天垓300能够在这些场景中取得性能领先，一个关键是针对实际负载进行了架构优化。
基于天枢架构的天垓300，单卡显存容量达到144GB，并围绕Attention、MoE等计算环节进行优化。长上下文需要容纳更多状态，也增加了Attention计算压力。MoE虽然通过激活部分专家控制计算量，却带来了小规模矩阵计算、专家分配和数据交换等新问题。
为此，天垓300让Tensor Core与Vector Core并行处理Attention中的矩阵运算和Softmax等向量操作，同时针对MoE中的小规模矩阵、低精度计算与通信协同进行优化。
这些针对性设计，有助于减少不同计算环节之间的等待，把更多硬件资源转化为实际产出。
此次披露的测试与现场演示，展现了天垓300进入高端推理实际竞争区间的能力。对客户而言，下一步要比较的，是在同样的响应要求下，系统能承接多少请求、持续运行表现如何，以及需要付出多少设备、电力和运维成本。
推理需求催生专用芯片，GPU的价值会被削弱吗？
随着推理需求快速增加，以LPU为代表的专用推理芯片备受关注，不同计算环节对硬件的要求也进一步分化。
Prefill处理输入，通常更偏向计算密集型任务；Decode逐步生成Token，更容易受到访存、任务启动和同步开销的影响。进一步拆开，Attention与FFN对计算、带宽和存储的需求也不相同。
当某类任务足够稳定、规模足够大，专用芯片就有机会通过针对性设计，获得更好的性能或能效。让一种通用GPU处理所有环节，未必始终对应最低的成本。
天数智芯展示的AFD分离方案，就是对这种分工的回应。
它在Prefill与Decode分离的基础上，继续拆分Decode阶段的Attention与FFN：天垓300承担Prefill全流程计算，以及Decode中的Attention；合作厂商的XPU承担FFN专家网络计算。
运行过程中，GPU把中间激活数据发送给XPU，XPU完成计算后返回结果，继续推进模型生成。天数智芯给出的对比结果显示，在所展示的方案中，Decode推理延迟降低60%。
天数智芯AI与加速计算负责人单天逸表示：“未来的AI算力系统，不一定由一种芯片包打天下。关键是如何把它们高效组织起来，让整个系统发挥更大的价值。”
异构分工为不同芯片提供了合作空间，GPU的通用性则关系到系统能否持续适应模型变化。
天数智芯副总裁邹翾对雷峰网表示，专用硬件可以在固定工作负载中实现较大的性能提升，但模型与算法仍在快速迭代，公司坚持以通用框架为基础，结合专用加速能力。
PPTV创始人再创业做“算力版滴滴”，日均消耗token超万亿，却难赚到钱？
（本文作者为 创业最前线，钛媒体经授权发布）
文 | 创业最前线，作者 | 于莹，编辑 | 王亚静
当AI大模型开始收费，卖Token的“工厂”逐渐成为市场最“吸金”的AI赛道之一。
在海外，被黄仁勋称为“AI工厂的台积电”的Fireworks AI7月拿到15.05亿美元融资，估值升至175亿美元；在国内，10月12日，上海AI Infra（人工智能基础设施）企业无问芯穹确认已完成23亿元C轮融资，累计融资额达43亿元。
资本追捧之下，做Token“工厂”的企业，开始走向大众视野，硅基流动、派想未来等企业纷纷向港交所递交上市申请。
其中，派想未来的创始人是曾创立了PPTV的“80后”学霸姚欣，他在退出ppvt后，在AI算力爆发前悄悄布局了算力赛道。
派想未来是一家独立分布式云计算服务商，业务始于边缘云计算服务领域，2025年营收已经达到7.7亿元。按灼识咨询口径，公司是中国最大的独立AI云计算服务商。
不过，公司至今尚未盈利，经营现金流也呈现净流出状态。而在可转股的条件压力下，公司上市需求十分迫切。
只是，这门在资本市场陆续拿到“大结果”的Token生意，底色究竟如何呢？
1、打造token工厂，要做“算力版滴滴”
通俗来说，Token工厂就是专门规模化生产推理服务的“车间”，把电力、芯片、模型和数据，持续转化为可计量、可调度、可定价的模型调用能力。
而一般的传统云计算巨头，主要是“中心化大厂”逻辑，在全国建几个超级数据中心，通过长途光纤把算力输送给用户。这种模式，在需要超低时延、海量并发的5G和智能化时代，容易遇到物理瓶颈。
与阿里云、腾讯云等公有云平台不同，派想未来并非自建数据中心，而是通过“蜂群思维”搭建起中国最大的分布式算力网络。
公司发现，社会上存在大量分散、闲置、异构的第三方闲置算力资源，海量的算力需求与分散且尚未充分利用的算力供给之间存在重大错配。
所以，公司搭建了算力调度平台，把它们当成“私家车”，汇聚和管理异构资源的闲置算力，以最优方式向客户部署，从而加速内容传输及提升低延迟计算体验。
在接受采访时，创始人姚欣将公司比作“算力版滴滴”，称公司不采购、不持有大规模GPU服务器与数据中心，不自主搭建自有智算园区，全部算力资源向外采购整合，再通过自研分布式调度系统匹配供需。
（图 / 招股书）
可以说，派想未来的核心资产不是硬件，而是一套极度聪明的分布式云计算架构和统一调度平台。正是通过这种调度能力，公司把全国1340多个县市的4600多个闲置节点盘活，成功构建起了一个边缘算力网络，把算力直接下沉到网络的最边缘。
（图 / 招股书）
这种“轻资产、重调度”的模式，在2023年撞上了大模型爆发的超级风口。招股书显示，公司通过分布式调度后，GPU利用率在2025年一直超过75%，远高于行业40%-50%的平均水平，模型推理成本理论上可以下降超过80%。当AI时代来临的时候，这种优势被放大。
（图 / 招股书）
而打造派想未来，并非是姚欣的首次创业。
2004年，正在华中科技大学读研一的姚欣，为了解决同学们在宿舍看球赛直播卡顿的痛点，敏锐地捕捉到了P2P技术“人越多越快”的特性，索性休学创办了PPTV的前身PPLive。
在那个宽带和分发基础设施极度匮乏的互联网黄金年代，PPTV一度坐拥4.5亿全球用户，成了仅次于优酷的全国第二大视频平台，2014年PPTV最终被苏宁集团收购。
2018年，姚欣与当年的老搭档、华科学弟王闻宇再度联手，成立了派欧云计算（上海）有限公司（以下简称派欧云，派想未来在中国的合并联属实体之一），如今一起带领着派想未来冲刺资本市场。
招股书显示，IPO前，姚欣及其配偶吕姗姗，通过数家实体合计控制约48.05%的股份投票权；王闻宇间接持股10.17%，其他投资者包括Millennium、IMO Ventures、CCV Fund、华业天成、磐霖、沸点等。
2、营收2年翻倍，但为何越卖越亏？
目前，派想未来的商业模式采用云计算的收费模式，即按算力使用量或使用时长计费，主要付费客户包括大型互联网企业，以及中小型AI创业企业。
2023年至2025年（以下简称报告期内），公司收入从3.58亿元迅速增长至7.70亿元，复合年增长率达46.6%。
其中，边缘云计算服务收入（包括边缘节点服务、边缘CDN）分别为3.58亿元、5.48亿元、6.51亿元，长期占据总收入84%以上。
而AI云计算服务收入从2023年的26.5万元暴涨至1.19亿元，2025年同比增长超过10倍，占比提升至15.5%，成为招股书重点渲染的第二增长曲线。
（图 / 招股书）
营收高增，毛利率却明显下滑。报告期内，其综合毛利率从2023年的17.7%滑落到2024年的12.3%，再到2025年的9.4%，近乎腰斩。
（图 / 招股书）
在这背后，公司一直深陷亏损之中。报告期内，其净亏损分别为1.89亿元、2.94亿元和2.23亿元。如果剔掉可转换可赎回优先股公允价值变动等非现金项，经调整净亏损仍从0.37亿元扩大到1.05亿元，陷入了“卖得越多，亏得越多”的境地。
（图 / 招股书）
造成亏损的根源，主要出在公司快速增长的AI云计算业务上。
2025年，其AI云计算毛利率为-10.7%，而2023年、2024年则分别为-91.3%和-95.1%，虽然毛利率亏损收窄，但至今仍是一笔“亏本买卖”，而其收入占比越来越高，拖累了整体毛利率表现。
「创业最前线」了解到，公司边缘云业务，主要是把算力放到离用户最近的地方，专门给视频、直播、文件下载做加速（也就是边缘CDN）。
而AI云业务，则提供的是GPU算力和现成大模型接口（MaaS），客户不用自己买显卡、搭环境就能跑AI推理。但这块收入，主要来源于模型API，其为客户调大模型、按Token收费。
但Token价格并非为派想未来所定，而是模型官方定价。根据爱分析的报道，派欧云的定价页面和DeepSeek、GLM、Kimi等官方API文档定价一致。
（图 / 爱分析（上）、招股书（下））
原因在于，定价权在模型厂商手里。Token API是同质化商品，用户切换Token供给方的成本、门槛极低。因此，即使大模型厂商的推理算力供给不足，派想未来也难以溢价收费。
而在成本端，派想未来的GPU算力多是租赁而来，2025年GPU资源成本1.17亿元，几乎追平了整体的AI云收入（1.19亿元）。可以说，公司每多提供一个Token，就要多付一笔GPU租金。
（图 / 招股书）
从这点看，虽然公司把自身比喻成“算力版滴滴”，但滴滴的抽佣定价权在自己手里，而派想未来上游是掌握定价权的模型厂商，下游是可以一键切换的客户，两头都被挤压。
不过，需求端确实是十分紧俏，公司披露的日均Token消耗早已破万亿，说明市场确实存在广泛需求，只是需求旺盛并不等于赚钱能力，公司的利润空间尚未打开。
在招股书中，公司也提示：“随着对公司产品及服务的使用量增加，客户可能要求降低定价，部分客户可能出于商业、合规或其他原因选择使用其他云服务提供商。为此，公司为维持或增加收入，或需要在销售及营销方面投入更多的支出，这些都会对公司的毛利率产生不利影响。”
可见，公司的低毛利率并非一时阵痛，是这套“租卡—转售Token”模式的天然限制。另外，算力中介赛道门槛低、同质化严重，派想未来若想突围并不容易。
3、上市倒计时，仅剩两个月时间
值得注意的是，公司的经营命脉，压在极少数大客户身上。
2023至2025年，公司的前五大客户收入占总营收比重分别达92.5%、89.5%、79.0%，营收高度依赖大客户。
（图 / 招股书）
面对大客户，派想未来的回款周期较长，导致公司经营现金流持续为负。
招股书显示，公司通常向边缘云计算服务客户授予一至三个月的信贷期，2023年至2025年，贸易应收款项分别为0.9亿元、1.5亿元、1.68亿元。
同期，贸易应收账款周转天数分别为83天、79天及75天。报告期内，贸易应收款项中分别有93.6%、88.1%及74.9%为应收五大客户款项。
而从客户收取按使用量计算的费用，与向供应商付款之间通常存在时间差，从而导致潜在的经营现金流错配。报告期内，其经营现金流净额分别为-0.35亿元、-0.41亿元、-1.36亿元。
（图 / 招股书）
而公司运营颇为依赖融资。2023年至2025年，其累计股权融资额分别为0.6亿美元、0.6亿美元及1.06亿美元，计息银行借款分别为0.44亿元、0.74亿元及0.81亿元，逐年攀升。
（图 / 招股书）
此次港股IPO，公司拟将募资用于，提升公司于分布式云计算服务的技术能力及升级AI云计算服务；海外扩张，特别是在成熟市场及潜力尚未开发的新兴市场；战略投资及收购；营运资金及一般企业用途。
除了营运资金的压力，目前派想未来着急上市的原因，还有来自于悬在头顶的“赎回压力”。
招股书显示，派想未来的背后站着蓝驰创投、创世伙伴、华业天成等投资机构，以及早期重仓的百川智能王小川、前微软Azure中国区总裁申元庆、元气森林唐彬森。
从2021年天使轮后的4600万美元估值，到2025年B轮融资后的4.69亿美元，再到2026年5月底6月初刚完成的B-2轮，派想未来的投后估值已经达到5.76亿美元。
融来的钱虽然撑起了业务扩张，但也可能变成到期要还的债。当初为它背书的投资天团，此刻也正是手握赎回权的人。
2023年至2025年，公司分别录得可转换可
一夜之间，GPT-6.2 Astra 和 Fable 5.5 双双曝光，AI 圈疯狂 11 月要来了
刚进入 11 月，AI 圈又开始热闹起来了。
过去十几个小时里，Anthropic 和 OpenAI 接连传出新模型消息。有人在 Anthropic 官网源代码中发现了 Claude Mythos 5.5 的名字，还有开发者晒出疑似 Claude Fable 5.5 生成的 3D 游戏和交互作品。
另一边，OpenAI 尚未官宣的 GPT-6.2 Astra，也出现在部分用户的 Codex 使用统计和 API 响应记录中，甚至有人声称已经成功调用。
两家公司的新模型线索几乎同时出现，结合此前关于 Qwen4、GLM-5.5 和 Hy4 的传闻，也让 AI 社区开始猜测，11 月是否又会迎来一轮旗舰模型集中更新。
Claude 5.5 官网意外曝光，新模型要来了？
事情要从 Anthropic 官网的一次意外发现说起。
11 月 1 日深夜，科技博主 @kimmonismus 在 X 平台发帖称，Anthropic 官方网站的 HTML 源代码中，出现了尚未正式公布的 Claude Mythos 5.5。
根据截图，Anthropic 的 Mythos 产品页面中，隐藏的 H1 标题以及图片的 alt 属性，都包含 Mythos 5.5 字样。
随后，科技媒体 TestingCatalog 也关注到这一变化，并推测 Anthropic 可能正在筹备新一代模型。
对于不熟悉 Anthropic 产品线的读者，可以简单理解为，Mythos 主要面向网络安全和生物学等高风险研究领域，访问权限受到严格限制；Fable 则在相同底层模型基础上加入安全防护，面向更广泛的用户提供服务。
因此，Mythos 5.5 的名称意外现身官网，也让外界开始猜测，面向普通用户的 Fable 5.5 是否即将登场。
值得注意的是，今天凌晨，开发者 @chetaslua 进一步发现，Mythos 5.5 曾在官网 HTML 中出现六次，但正文介绍仍然停留在 Mythos 5.1。
向左滑动查看更多内容
随后，另一位用户 @pankajkumar_dev 表示，Anthropic 已经删除了涉及 Mythos 5.5 的网页内容。
与此同时，另一条关于 Fable 5.5 的线索也被翻了出来。
早在 10 月 13 日，开源大模型 API 网关 LiteLLM 的 Azure 价格配置中，就曾出现过 azure_ai/claude-fable-5-5 的模型标识。相关配置随后被移除，但第三方 MLflow 模型目录仍保留了对应记录。
根据流传的配置信息，Fable 5.5 可能支持 100 万 token 输入上下文、12.8 万 token 最大输出，每百万输入 token 为 10 美元，输出为 50 美元，缓存读取价格则为 0.25 美元。
至于新模型究竟什么时候发布，社区同样众说纷纭，有人预测本周，有人预测下周，也有人认为要等到 11 月中旬，因为这可能是 Anthropic 上市的最佳献礼。
Claude Fable 5.5 编程能力曝光，3D 游戏都能一句话生成？
相比官网源代码里的几个字符，更让开发者兴奋的，是 X 平台上接连出现的 Fable 5.5 演示视频。
11 月 1 日，开发者 @notjazii 发布了一段约 36 秒的对比视频，声称使用相同提示词，让 Claude Opus 5.5 和 Fable 5.5 分别生成 3D 场景。
画面中，Opus 5.5 主要呈现湖泊、公园、喷泉以及四季变化的自然景观，整体更接近微缩模型展示。而 Fable 5.5 一侧则出现了农场四季变化、城市建筑逐渐生长、港湾昼夜交替，甚至还有火山喷发、火箭升空等动态事件。
相比静态场景，Fable 5.5 的作品更强调连续动画、时间变化和物体状态。
发布者认为，新模型在此次对比中全面胜出。
紧接着，他又展示了一款据称由 Fable 5.5 生成的 Minecraft 风格游戏。
在约 92 秒的视频中，玩家能够以第一人称视角探索方块世界，穿过森林、洞穴、海岸和沙漠河流，甚至还能看到熔岩区域、粉色树冠森林，以及水面上的动态光照。
游戏界面同样相当完整，包括镐子、物品快捷栏和分类背包等常见元素。作者声称，整个作品只用了一个提示词，花费一到两个小时便完成生成。
另外两段演示也颇有看点。
开发者 @abhinavflac 展示了一艘据称由 Fable 5.5 通过 zero-shot 提示词生成的交互式历史帆船。
镜头能够围绕船身移动，观察甲板、炮门、帆面和绳索结构，海浪、船尾尾流以及夕阳下的水面反光也被呈现出来。
另一位开发者 @chetaslua 则展示了一款基于 Three.js 的跑车应用，用户可以从不同角度观察车辆，还能随时切换红、蓝、银、黑和荧光绿等车漆颜色。
从这些作品来看，开发者已经开始尝试让 AI 承担更复杂的 3D 建模、场景动画和交互开发任务。不过，相关视频并未附带完整提示词、调用日志和源代码，模型身份、生成耗时以及是否一次完成，仍主要依赖作者自述。
倘若后续能够得到验证，Fable 5.5 在复杂交互应用开发方面的表现，确实值得期待。
GPT-6.2 Astra 也来了，Codex 用户提前用上新模型？
Anthropic 这边消息不断，OpenAI 也出现了新的动静。
11 月 1 日，X 用户 @TokenGremlin 发帖称，部分用户的 Codex 使用统计中，出现了一个尚未正式公布的模型名称。
gpt-6.2-astra
根据网友的截图，在某个账户的模型使用占比中，GPT-6 Astra 占据约 80%，GPT-6.2 Astra 则占据 19.8%。
有意思的是，用户在模型选择界面看到的仍然是 GPT-6 Astra，但后台统计却显示了新的模型标识。
随后，开发者 @Mr_Salio 也晒出一段 API 响应数据，其中 response.model 字段显示为 gpt-6.2-astra。
两条线索分别来自使用统计和响应记录，让外界猜测 OpenAI 可能已经开始测试新版本 Astra。
更进一步，今天凌晨，用户 @ai_funss 表示，升级 Codex CLI 后，可以通过终端指定模型名称，尝试调用 GPT-6.2 Astra。
其展示的截图不仅出现了 gpt-6.2-astra medium 标识，还显示 1M window，也就是 100 万 token 上下文窗口。
至于性能，社区内已经出现两种截然不同的评价。
@TokenGremlin 认为，GPT-6.2 Astra 相比前代有明显进步，尤其是在设计、技术能力和编程表现方面，部分任务甚至可能超过 Claude Opus 5.5。
但另一位用户 @Da7_Tech 却认为，新 Astra 尚未达到 Opus 5.5 的水平，更不用说传闻中的 Fable 5.5。
值得一提的是，10 月 31 日，OpenAI 官方账号还发布了一段视频，配文「未来有很多种可能。你想创造哪一种？」，也引发了外界对其是否由新一代模型生成的猜测。
回顾最近一个月，全球大模型厂商的更新节奏明显加快。
10 月 28 日，Google DeepMind 刚刚发布 Gemini 4 Carbon，如今 Anthropic 和 OpenAI 又相继传出下一代模型的线索，国内的 Qwen、GLM、Hy4 等模型也被寄予新版本发布的期待。
如今的大模型竞争，已经逐渐进入一个有些荒诞的阶段。
一款旗舰模型还没正式亮相，网友就已经通过官网源代码、后台记录和演示视频，找到了各种蛛丝马迹，甚至提前争论起谁才是下一任世界第一。至于这些线索究竟是厂商无意留下的痕迹，还是有意放出的预热信号，恐怕只有厂商自己清楚。
但流量本身无疑是一把双刃剑。毕竟，AI 厂商最难击败的对手，可能已经不再是彼此，而是被上一代模型不断抬高胃口的用户。
当真实产品终于登场，如何超越互联网提前为它塑造的想象，或许比击败竞争对手更加困难。
#欢迎关注爱范儿官方微信公众号：爱范儿（微信号：ifanr），更多精彩内容第一时间为您奉上。
GitHub Copilot Fusion
GitHub's AI coding agent that picks and combines models
Tech stocks just lost more than $1 trillion in a day—a gothic horror test for the ‘moats’ Warren Buffett taught Wall Street to love
Markets are ushering in a particularly gothic Halloween as the AI trade exposes a crumbling aristocratic pile of early 21st century capitalism. Wall Street has been describing itself in castle-adjacent terms for years.
Jamie Dimon has long talked to JPMorgan shareholders about the bank’s “fortress balance sheet.” His 2015 letter set aside a section to answer the question of what the phrase even meant. By 2021, he’d added a second castle word, warning that “the moats that protect this company are not particularly deep.”
The moat came from Warren Buffett. In a 1999 Fortune article with Carol Loomis, he argued that the key to investing is a company’s competitive advantage and how durable it is, and that the businesses worth owning have wide moats around them.
The gothic novel’s favorite picture of danger is a castle falling apart, and for about 250 years the two have traded images: the castle and the moat, the vampire, the creature built in a lab. This Halloween season has them rushing to the front of the conversation, with a two-day rout in tech stocks on Thursday and Friday turning on whether the AI giants’ moats will hold, as gothic horror booms on bookshelves and movie screens.
Horror took 15.5% of the domestic box office in 2025, according to The Numbers, and U.S. horror book sales rose 7% through the first nine months of 2025, per Circana BookScan, while adult fiction overall grew just 1%. The New York Times asked Friday why every book is gothic now, with publishers putting the label on everything from family sagas to space operas, and several film critics commenting that Nathan Fielder’s smash-hit business-scandal documentaryYou Can See Everything is a twisted, even gothic romance in the tradition of Alfred Hitchcock.
The moats get tested
The selling started with Alphabet. The stock fell 12.6% Thursday, wiping out roughly half a trillion dollars in market value, after the Google parent’s quarterly report. Revenue from the core search unit fell short, partly because of the popularity of less-profitable “AI mode” searches and the company burning cash faster than expected. Its free cash flow came in at negative $9.1 billion, against the negative $2.23 billion analysts expected, per FactSet.
The S&P 500’s worst performer was software giant ServiceNow, which dropped 27% the same day on slowing revenue growth and disappointing guidance. By Friday morning, it was Amazon’s and Intel’s turn, with Intel down roughly 25% before the opening bell. Big Tech lost nearly $900 billion in market value after another sharp rise in AI spending in the third quarter, the Financial Times calculated. Nasdaq 100 companies lost more than $1 trillion in market value Friday alone. Reddit, whose shares fell more than 22% premarket Friday, is pulling up the drawbridge: it’s moving to a logged-in experience as AI eats search.
The irony is spooking investors. AI is eating into the businesses that throw off the cash, and the cash is going into AI anyway. The hyperscalers have raised $326 billion in stock and bonds this year to pay for the build-out, more than triple 2025, according to S&P Global Market Intelligence data.
The question underneath it is the one Apollo chief economist Torsten Slok put to clients the morning Alphabet fell. “With AI disruption potentially around the corner, there is a lot of discussion about moats at the moment,” he wrote in his Daily Spark note Thursday.
He built on finance professor Aswath Damodaran, the “dean of valuation,” who compiled a list of sources of competitive advantage, including brand strength, switching costs, network effects, cost advantages and legal protection. Then he boiled it down to an AI moat checklist of three questions: does the company own data its rivals can’t get, is its product part of how customers do their jobs every day, and can it charge more without customers walking away?
Judging by that third test, the companies selling AI don’t seem to measure up. OpenAI’s new GPT-6.1 Terra model costs up to half as much per token as the version before it. An index from expense-management firm Ramp, which looked at what companies actually pay Anthropic and OpenAI, shows token prices down roughly 35% since the end of June, in what Axios called a price war.
The companies paying for AI aren’t getting much back yet, either. In an Oct. 24 note, Slok sent a tremor through Wall Street terminals with an argument that AI “has delivered zero financial payoff to corporate America outside of tech.” Revenue per worker and profit margins are rising for the Magnificent Seven and flat for the rest of the S&P 500, he noted, and mainstream companies are starting to treat token spending as a cost center that weighs on margins rather than a source of growth.
The first castle was a fake
Xavier Aldana Reyes, a researcher at the Manchester Centre for Gothic Studies, told the Times that interest in the gothic tends to surge during periods of social anxiety and political turmoil. Literary agent Mary C. Moore put the genre’s subject more bluntly, telling the paper that “Gothics are all about exposing what is rotten at the core of something shiny/wealthy.”
The genre was invented by a scion of wealth who knew the old order from inside and drew a salary without working at all. Horace Walpole was the son of UK Prime Minister Robert Walpole, and hisThe Castle of Otranto, published in 1764, is generally regarded as the first gothic novel.
He passed it off as a translation of an obscure Italian work, found in the library of an old Catholic family in northern Eng
AI boom hits a wall on Wall Street as OpenAI cuts fundraising target while Meta gets a credit warning over its cash flow, spending and monetization
The seemingly endless gusher of capital that’s been showered on Silicon Valley during the AI boom may be coming to an end soon as Wall Street pushes back on massive spending plans.
Following a brutal selloff in tech stocks over the past week, OpenAI slashed the target on its latest fundraising round to $20 billion from $30 billion, sources told the Financial Times. That also cut its expected valuation to $1.25 trillion from $1.4 trillion.
The funding round was also upstaged by Anthropic, according to the report, as OpenAI’s arch-rival prepares to go public next month. Ahead of its IPO, Anthropic filed an S-1 on Monday that showed revenue has soared and its cash flow turned positive in the first half of this year.
OpenAI didn’t immediately respond to a request for comment.
The timing couldn’t have been worse. Quarterly earnings reports from hyperscalers this week spooked investors. Alphabet saw its cash flow sink deeper into negative territory in the third quarter, with the Google parent also joining Meta and Amazon in charging ahead with massive spending plans. On Thursday, Alphabet fell about 12%, its worst day on record.
To be sure, there were other positive signs as well, in addition to Anthropic’s bullish S-1. Microsoft reported strong cloud growth and stayed cash flow positive, though it said its capital spending will accelerate, while AI chip maker SK Hynix blew away earnings forecasts.
But the narrative had already turned bearish, especially after AI debt fears sent shares and bonds of “neocloud” companies tumbling earlier this month.
At the same time, OpenAI was seeking more capital just as other AI giants tapped financial markets heavily. This past week alone, Elon Musk said SpaceX closed $40 billion in debt financing, Meta announced plans to raise $95 billion in equity, and Microsoft raised $25 billion in a bond sale.
The prospect of hyperscalers continuing to hemorrhage funds was too much for S&P Global, which cut its outlook on Meta to “negative” from “stable” late Friday, citing negative cash flow and uncertainty around capital expenditure plans.
The credit ratings agency maintained Meta at AA- but warned the social media leader’s leverage could top 1.0x as soon as 2027, which is the downgrade threshold for the current rating. That’s up from about 0.1x in 2025 and primarily reflects its strategy of leasing data center capacity rather than from its debt issuance.
Meanwhile, S&P sees Meta’s capex reaching $140 billion this year, topping $225 billion in 2027, hitting $252 billion in 2028, and then flattening out from there. As a result, it expects free operating cash flow to remain negative: -$22 billion this year, -$71 billion in 2027, and -$39 billion in 2028.
Of course, other hyperscalers are spending heavily and running cash deficits, but S&P also said Meta’s ability to monetize its AI investments is more limited than that of its peers.
“While we believe the company intends to monetize these investments via improved advertising tools that support faster ad revenue growth, as well as from the development of new revenue streams stemming from its introduction of agentic apps such as Muse, it is unclear when these revenue streams will generate material contributions that exceed their rising costs over the next few years,” S&P said.
Meta’s rivals, however, are monetizing their data centers by leasing compute capacity, including to frontier model labs and enterprise customers, it added.
In a statement to Fortune, a Meta spokesperson said the company disagrees with S&P’s assessment of its AI monetization capabilities, pointing to “significant product and business opportunities” ahead.
“Meta is already delivering stronger engagement, better recommendations, and greater returns for businesses as a result of AI investments,” the spokesperson continued. “We believe the technology and infrastructure we’re building now will open new revenue streams, create long-term shareholder value, and bring us closer to delivering personal superintelligence to everyone.”
The bearish environment on Wall Street also comes as rising bond yields have sent borrowing costs sharply higher, the political backlash against data centers is expected to help Democrats retake Congress, and the Iran war could re-escalate after midterm elections.
Still, some bullish voices aren’t convinced the AI boom is about to go bust, and even think some retrenchment is a good thing.
According to market veteran Ed Yardeni, the AI investment boom faces financial, physical, and political constraints. If any of these constraints worsen, stocks could dive further, especially considering that AI is now a major driver of corporate earnings and GDP growth, he said in a note this week.
“However, we view these constraints as healthy for the durability of the AI boom!” he added.
For example, if the supply of capital doesn’t flow as easily, it could lead to more efficient allocation, lowering the risk of overinvestment. Yardeni expects hyperscalers to retain access to financing—thanks to their strong balance sheets—and keep spending.
In his view, it all comes down to the U.S. economy, which he said has performed remarkably well over the past few decades. Even the midterm elections are unlikely to derail it, regardless of which party controls Congress.
“Further AI jitters therefore could set up buying opportunities rather than signal the end of the boom,” he said.
This story was originally featured on Fortune.com
Regulate the Deployment, Not Just the Model
A who-does-what plan for AI safety, built around Trusted Deployers: certified hosts that keep dangerous models guarded without ending open access.
AI has reached a new level of public interest. That may seem like a silly thing to say, as it’s been of interest for a while. But prior (broadly public) interest was about it as a tool, or about the money to be made or lost, or about an effect on jobs.
The new interest demands policy. It demands action. The new interest is connected with risks I consider real. Being real doesn’t make these risks certain. But being real, even with uncertainty, is reason enough to act.
I’ve written lately about risks, cybersecurity risks, and loss-of-control risks. These tried to explain complex topics, starting from the initial premise. Recommendations follow the explanation. This piece will be a bit different. Its focus is upon what I think we should be doing. You might consider it a policy piece. Policy in a completed form has to be a bit broad. There’s always more than one thing going on. As such, I’ll lay out my outline first, and then offer below the extra detail, background and context that I think is important to understanding each point in that outline.
If you’re scanning the outline and see something you disagree with, I’d suggest you continue on, because the outline is, by design, abbreviated, and not fully descriptive.
This isn’t the only effort of this type. Institute for Progress (IFP) and AI Futures are two well known efforts. What distinguishes this most of all is thinking more deeply about how models are deployed. This is a background element that’s somewhat ignored, because it’s not as novel and is not as visible to users. But it’s absolutely critical. I don’t expect this plan would replace plans from well known organizations, but by illustrating a gap, I hope to inspire improvement amongst those plans.
Who and What
This outline is categorized by who would be expected to do what. In some cases, someone is expected to do it, but there’s also another role, of providing the incentives that improve the expectation they will.
Enforcement is deliberately vague. It depends on jurisdiction, has to fit existing legal frameworks, and will need to be worked out, and negotiated, through the political and legal systems. It would feel like I was putting the cart before the horse to even guess at how that would take shape at national and international levels.
Who does what. Model weights above the threshold reach users only through certified trusted deployers. Independent evaluators, certified by a regulator with legal authority, test the models and audit the deployers.
Layer 1: Infrastructure and Distribution
Trusted deployers
Who are they? A new designation that existing and new organizations can be certified as. Expect major cloud providers (AWS, Azure, Google), AI labs (for self-owned) and neoclouds to join immediately. The plumbing largely already exists, but needs standards and certification.
Revenue model? Charging for compute and operational services as a utility function. Or, an internal unit (for a lab, government, or large organization)
Core Responsibilities
- Develop secure environments to host models that receive an independent AI Deployment Safety Level (AIDSL) rating. Example, but not comprehensive list of requirements:
- Comprehensive monitoring, telemetry, and auditing (including multi-turn agent pattern detection).
- Resistant to external cyberattacks and insider threats.
- Resistant to internal cyberattacks. At minimum red-teamed by deployed model and a second highly capable model.
- Guardrails applied by default to all deployments which prevent access of dangerous capabilities.
- Strict controls for any guardrail exceptions as used for “limited” access deployments. For example, accessing users must be known directly, or an evidence chain validated using only trusted organizations.
- Understand the sensitivity of any model deployed. Never deploy a model in an environment with an AIDSL rating insufficient to the combination of deployment type and model sensitivity.
- Use internal qualification for small custom models and Independent Evaluator data for larger/frontier models.
- Models under training assumed to be at least X% more capable than the best model trained to date at the given scale. If the training scale is unprecedented, increase X by a size factor. (X should be established by independent evaluators)
- Perform or validate Know Your Customer (KYC) verification for all model providers, for Independent Evaluators and anyone with access to deployments with guardrail exceptions.
- Provide access to Independent Evaluators to confirm secure environments meet standards
<
Forecasting space weather risks on power grids
At a glance
- End-to-end forecasting: A machine learning pipeline uses forecast-time solar-wind information to generate location-specific risk estimates for 66,935 substations in the continental United States.
- Physics and place: The system combines Auroral Electrojet (AE) and Disturbance Storm Time (Dst) forecasts with local latitude, geology, and ground conductivity.
- Advance warning: The pipeline detected nearly 80% of major space-weather events during the evaluation period and can warn grid operators 30 to 60 minutes before a specific risk appears.
Forecasting a threat to critical infrastructure
During the May 2024 geomagnetic storm, utilities across North America prepared for possible impacts as auroras extended far beyond their usual range. The storm would degrade GPS accuracy and satellite operations, impacting farm operations in North America and Europe. Modern society depends on reliable electric power, yet extreme space-weather events can induce currents in transmission networks that damage equipment and increase operational risk. The challenge is not only knowing that a storm is approaching, but estimating when and where its effects could be most severe with enough warning for grid operators to respond.
During my summer internship at Microsoft Research, I developed a machine learning system that forecasts space-weather risk across 66,935 substations in the continental United States. The system measures how the sun’s activity can affect the earth’s magnetic field and, ultimately, the power grid. It combines solar-wind observations, forecasts of the Auroral Electrojet (AE) and Disturbance Storm Time (Dst) indices, physics-informed constraints, local geological conductivity, and grid-infrastructure data to produce location-specific risk estimates 30-60 minutes ahead of potential impact.
Building an end-to-end prediction pipeline
Space-weather prediction is difficult because it spans several coupled systems. The solar wind changes rapidly, its interaction with Earth’s magnetosphere is irregular, and the resulting ground effects depend on local conditions. Regions with resistive bedrock can experience stronger geomagnetically induced currents (GICs) than regions with more conductive geology. Transmission-line orientation, latitude, and other power-system characteristics further influence the exposure of individual assets.
The pipeline addresses this complexity in three stages, shown in Figure 1. First, solar-wind measurements from the L1 Lagrange point are used to generate forecasts of the AE and Dst indices, while geological conductivity and location features are assembled for each substation. Second, a gradient-boosting model combines these forecasted and location-specific inputs to estimate dB/dt, the rate of magnetic-field change associated with GIC risk. Finally, the resulting predictions are converted into location-specific risk estimates and aggregated into a continental risk assessment. A system of 50 AI agents helped explore features, validation strategies, and model configurations across the pipeline. Only public data sources were used, included NASA OMNI and NASA-aggregated Kyoto World Data Center data, INTERMAGNET and U.S. Geological Survey magnetometer observations, and GridSFM-derived grid data.
Evaluating geomagnetic forecasts and infrastructure risk
The AE predictor was designed to forecast rare, high-intensity geomagnetic activity that drives infrastructure risk. In the 2020-2026 evaluation period, the model produced forecasts spanning nearly the full observed range of AE activity and outperformed several empirical solar-wind-based approaches. The Dst predictor provided an additional signal describing large-scale geomagnetic storm strength. During the most geomagnetically active periods of the 2020-2026 evaluation period, the machine-learning model outperformed the Burton equation on 62.2% of individual hours. The model also produced a substantially wider prediction range than Burton-style approaches and improved severe-event detection in the end-to-end forecasting system by 1.2 percentage points when combined with AE forecasts.
| Predictor | Result | Baseline comparison | Impact |
|---|---|---|---|
| AE | 410.2 nT RMSE | Lower RMSE than empirical and solar-wind-only baselines | Captures extreme-event activity used by the GIC forecasting system |
| Dst | 7.2 nT RMSE | Outperformed Burton on 62.2% of peak-activity hours and provided wider prediction range | +1.2pp severe storm detection when integrated into final GIC system |
The GIC risk stage was evaluated differently. There is no equivalent widely deployed operational system that provides a direct industry benchmark for this calculation, so the machine learning model was compared with simple linear regression. As Figure 2 shows, the system achieved detection rates of 76.5% for major events (≥10 nT/min), 81.2% for severe events (≥20 nT/min), and 64.1% for extreme events (≥50 nT/min). False-alarm rates increased with storm severity, reflecting the trade-off between missed events and cautious alerts. Performance varied by latitude, with the highest detection rates at northern stations where geomagnetic activity is strongest.
Replace Microsoft 365 with a $27.97 lifetime Office license for Windows
TL;DR: Get Word, Excel, PowerPoint, Access, and more with a Microsoft Office Professional 2021 lifetime license for Windows, now $27.97 (reg. $219.99).
A Microsoft 365 subscription bills you every year for apps most people open every day. Functionally, that means you’re stuck paying for the same products forever. If you’d rather pay once, try a Microsoft Office Professional 2021 lifetime license on sale for $27.97 (reg. $219.99).
Your one-time purchase gets you Word, Excel, PowerPoint, Outlook, OneNote, and Access, plus the free version of Teams. Access comes with the Professional edition, so anyone who works with databases gets that tool alongside the usual documents, spreadsheets, and slides.
You can install your apps on one Windows 10 or 11 PC that you use at home or at work. Your license will be tied to that device instead of a Microsoft account. This version won’t run on Windows 7 or 8, and it doesn’t work in Parallels or other virtual machines, so check your setup before you buy.
After checkout, your redemption code arrives by email, and you have seven days to redeem it. Unredeemed licenses can go back for store credit within that same window.
Stop paying every year for the Office apps you use every day.
Through Nov. 1 at 11:59 p.m. PT, you can get a Microsoft Office Pro lifetime license for just $27.97.
Microsoft Office Professional 2021 for Windows: Lifetime LicenseSee Deal
StackSocial prices subject to change.
Windows Search keeps lagging, and there's a built-in fix most people never try
There is no lack of frustrating and unwanted features on Windows, and Search is perhaps one of the worst offenders. Your experience with Windows Search has probably revolved around irrelevant results, missing files, and painfully slow performance. While Microsoft has improved relevancy and removed ads from Search, using it can still feel unusually slow, especially on a computer you've had for a while.
Microsoft offers a quarter of its workforce voluntary severance while shielding AI-critical teams
Microsoft is offering voluntary severance packages to about 60,000 employees, roughly a quarter of its workforce, in the second such program this year. The company's headcount has held flat at around 223,000 since 2022 even as it plans to spend about $170 billion on AI infrastructure in 2026 alone.
The article Microsoft offers a quarter of its workforce voluntary severance while shielding AI-critical teams appeared first on The Decoder.
Workflow vs agent
Workflow vs Agent: Deterministic Orchestration & Dynamic Tool Selection
Quick Diagnosis
| Situation | Recommended approach | Why it fits |
|---|---|---|
| All steps known ahead of time (e.g., fetch data → prompt → parse → store) | Workflow | Immutable execution, easy testing, predictable token usage |
| Path depends on content or user intent (e.g., “answer any compliance question”) | Agent | Chooses tools at runtime, handles unknown sub‑tasks |
| Regulatory audit required (immutable logs, version‑controlled prompts) | Workflow | Deterministic runs produce repeatable audit trails |
| Live numeric value needed (credit score, market rate) | Tool call from workflow (authoritative API) | Guarantees freshness; does not rely on stale retrieval |
| Batch processing with tight cost budget | Workflow | One LLM call per step, predictable token usage |
| Real‑time, ad‑hoc question answering | Agent | Dynamic reasoning, on‑the‑fly tool selection |
| Mix of known skeleton + occasional live‑value request | Hybrid (workflow + runtime API/agent) | Keeps core deterministic, adds flexibility only where required |
Decision Flow
flowchart TD
A["Is the full execution path known ahead of time?"]
B["Can every required tool be enumerated?"]
C["Use a deterministic workflow"]
D["Is the request likely to need unknown sub‑tasks?"]
E["Add an agent layer for dynamic tool selection"]
F["Can live values be fetched via an API?"]
G["Expose API call from the workflow"]
H["Leave decision to the agent at runtime"]
A -->|Yes| B
B -->|Yes| C
B -->|No| D
D -->|Yes| E
D -->|No| F
F -->|Yes| G
F -->|No| H
Architectural Responsibilities
| Layer | Core duty | Typical implementation |
|---|---|---|
| Deterministic Orchestrator | Define the fixed sequence, validate inputs, enforce business rules, emit immutable logs | Azure Logic Apps, Temporal, Airflow, hand‑rolled Python DAG |
| Agent Core | Detect intent, decide which tool or retrieval source to invoke, manage ReAct‑style loops, handle fall‑backs | LLM prompting with function‑calling schema, structured reasoning templates |
| Tool / Retrieval Layer | Provide static knowledge (document store) or fresh data (authoritative API) | Azure Cognitive Search, public REST endpoints, token‑protected credit‑bureau API |
| Governance & Auditing | Record every step, version‑control prompts, certify compliance checks | Write‑once log storage, signed manifests, CI linting of workflow definitions |
| Cost & Latency Monitor | Track token usage, external‑API count, per‑step latency | OpenTelemetry exporters, custom metric dashboards |
Python Decision Helper (Teaching‑Only)
# decision_helper.py – pure stdlib, no external packages
import json
from pathlib import Path
from typing import Literal, TypedDict
class DecisionInput(TypedDict):
"""Minimal data required to pick workflow vs. agent."""
steps_known: bool # True if the entire sequence can be enumerated
tools_enumerated: bool # True if every tool can be listed up‑front
live_value_needed: bool # True if an authoritative API must be called
dynamic_subtasks: bool # True if request may spawn unknown sub‑tasks
def choose_strategy(data: DecisionInput) -> Literal["workflow", "agent", "hybrid"]:
"""
Return the recommended execution style.
* `workflow` – fully deterministic pipeline
* `agent` – pure agentic control
* `hybrid` – deterministic skeleton with optional runtime API or agent
"""
# Basic validation – fail fast if the contract is broken
assert isinstance(data["steps_known"], bool), "steps_known must be bool"
assert isinstance(data["tools_enumerated"], bool), "tools_enumerated must be bool"
assert isinstance(data["live_value_needed"], bool), "live_value_needed must be bool"
assert isinstance(data["dynamic_subtasks"], bool), "dynamic_subtasks must be bool"
if 5 Key Reasons AV1 Will Play a Big Role in WebRTC Streaming
In this blog, you will learn why AV1 is getting closer to practical use in WebRTC-based real-time interactive streaming. We cover rapid progress in live AV1 encoding on GPUs and ASICs, widening browser and device support, and new WebRTC compatibility work like SVC and RTP payload formats. You will also see how Red5 Pro’s XDN and LCEVC can lower compute costs and expand device reach as AV1 adoption accelerates.
Advances supporting the next-generation AV1 codec have intensified over the past year, adding new urgency to the question of when the technology might become useful in WebRTC-based real-time interactive streaming applications.
There’s no question live video encoding supported by AV1 would benefit real-time streaming every bit as much as it stands to benefit traditional HTTP streaming. The codec achieves bitrate reductions in the range of 30%-40% compared to HEVC/H.265 and 60%-70% relative to AVC/H.264. Moreover, AV1 is available royalty-free with major tech sector backing through the Alliance for Open Media (AOMedia), whose members include Amazon, AMD, Apple, Cisco, Google, Hulu, IBM, LG, Meta, Microsoft, Netflix, Nvidia, Intel, IBM, Samsung, and many other leading players.
But the amount of computing power it takes to run the encoding and decoding processes has been a major impediment to AV1 adoption, especially when it comes to encoding at live video frame rates. As demonstrated in countless tests, there’s been no practical way to use AV1 to encode video in real time except at very low levels of resolution. For example, a series of recent tests conducted by cloud CDN and compute platform operator Cloudflare found that even if all 48 cores of the top-line AMD EPYC 7642 CPU processor were used to encode a 4K clip commonly used in such testing, it would take 43 seconds to encode two seconds.
But the perspective on the possibilities is shifting rapidly. And we at Red5 Pro are paying close attention to developments.
1. AV1 Live Encoding Is Closer than Ever to Commercial Viability
In 2021, Google released its desktop Chrome 90 browser with an AV1 encoder optimized for video conferencing with WebRTC. The encoder is targeted to two- to four-core CPUs for 2 Mbps transmission of 720p video at 30fps. This points to the fact that encoding AV1 for low bit-rate transmissions at sub-1080p levels is now doable on most PCs, which means the codec can be widely used to get decent-quality video over low-bandwidth mobile links that are the sole means of internet access in many parts of the world.
Cisco took another step in this direction in 2021 on the fringes of its Webex video conferencing service. The company now gives users the option to turn on live AV1 encoding to improve the quality of shared displays of video clips during conference calls.
Anyone on such a call with a device capable of decoding a low-bitrate AV1 stream will see the clip with much greater clarity. Cisco said it expects to expand the use of AV1 on Webex over time but had not announced any further steps as of YE 2022.
A much more dramatic sign of progress on the AV1 encoding front came in late 2022 with the introduction of powerful new AV1-enabled graphics engines by Intel, AMD, and Nvidia. As of this writing, for the first time, live AV1 encoding in fast-action gameplay can be executed by some versions of Intel’s Arc Alchemist and Nvidia’s RTX 4000 series running on gaming cards, which became commercially available in October. AMD released its far less expensive RX 7000 platform in early 2023 with no lofty claims for live AV1 encoding. Even so, early comparisons of the three tech companies’ latest advances, like this one reported by PC World, found them more or less on par with one another when it came to AV1 encoding of content for recording and archival usage.
It’s clear the AV1 live streaming barrier has been broken for gamers who want to shell out $1,500 or so to purchase the Intel or Nvidia gaming cards now on sale from multiple vendors. For example, testing of live video encoding with Intel Arc’s AV1 encoder against leading H.264 GPU encoders, as reported by PC World, found that when set to encode HD 1080p video at 3.5 Mbps, the quality of the video is superior to the output of H.264 encoders delivering the content at nearly twice that bitrate.
Of course, GPU cards significantly outperform encoders running on CPU cores or CPU cores enhanced with GPU acceleration, which remain the primary encoding engines in OTT streaming services. But these latest GPU platform advances point to the likelihood that the densification and architectural advances achieved with gaming encoders will be adopted in succeeding generations of hybrid CPU/GPU processors.
Another hardware advance signaling a role for live AV1 encoding occurred at the 2022 IBC trade show in Amsterdam, where RealSprint, the Swedish company operating the Vindral CDN service, revealed it would be introducing AV1 live encoding on its platform using ASIC (application-specific integrated circuit) hardware technology from NETINT Technologies. NETINT’s low-latency 4K AV1 encoder harbingers the emergence of ASICs as a potential game changer in advanced codec operations, according to NETINT COO Alex Liu, as quoted by Streaming Media. “[These applications] can only be met with ASIC hardware encoding,” Liu said.
2. Browser, Player Support for AV1 Decoding Is Gaining Wider Traction
Meanwhile, on the decoding side, there’s been a surge in browser and app support for AV1 software decoding on high-power devices. For now, this means that entities streaming on-demand content can use AV1 encoding with high confidence that a growing share of users will be able to take advantage of the bandwidth savings, which are beneficial to distributors as well as receivers.
The Google-backed Chrome browser and apps based on Android 10 or later can support software-based AV1 decoding on any device with enough processing power to accommodate the codec. Firefox, too