https://t.me/AI_News_CN
📈主流AI服务状态页通知 | 🆕汇集全网ChatGPT/AI新闻 #AI #ChatGPT
🔙备用群 https://t.me/gpt345
✨BEST AI中转 https://api.oaibest.com 2.8折起 支持OpenAI, Claude code, Gemini,Grok, Deepseek, Midjourney, 文件上传分析

Buy ads: https://telega.io/c/AI_News_CN
大模型真的能赚钱吗?AnthropicIPO文件给出首份财务破局答案

近日,Anthropic提交的IPO文件首次系统披露了前沿AI实验室的财务表现,直面解答了“AI公司能否实现盈利”的行业核心疑问。数据显示,该公司业务在规模化效应下展现出极高效率,其2026年第二季度营收达到116亿美元,同比大幅增长1368%,实现营业利润9.74亿美元,调整后营业利润率达19%,标志着前沿AI实验室的单位经济效益已正式转正。

在成本与客户结构方面,Anthropic的收入增速显著超越核心计算成本。2026年第二季度,其每投入1美元计算资源可产生2.08亿美元收入,远高于2025年第一季度的40美分。同时,公司客户集中度呈现优化态势,截至2026年9月,年收入超10万美元的客户已超过11900家,且前五大客户营收占比由一季度的36%降至三季度的21%左右。

尽管财务与现金储备表现稳健,IPO文件也揭示了潜在风险,指出公司面临近8000亿美元的计算基础设施协议敞口,但其中大部分为非固定承诺或与产能交付挂钩。这一进展证明了AI商业化变现的可行性,为正处于高投入周期的全球大模型产业提供了重要的财务参考样本。

via AI新闻资讯 (author: AI Base)
Kimi K3.1 Pro 对比 Claude Opus 5.5:同一像素画提示词实测

GMI Cloud 用同一提示词测试 Kimi K3.1 Pro 与 Claude Opus 5.5 生成精细的日式像素画街角,结论是 Kimi 更便宜但质量尚未达到 Opus 5.5 水平。

内容详情 · 原文链接

via AI情报局 | 全球 AI 动态简报 - Telegram Channel
电脑操作提速超 3 倍、API价格直接腰斩:OpenAI重磅发布GPT-6.1 Terra

近日,OpenAI正式推出了全新一代人工智能模型GPT-6.1Terra。该模型将核心优化方向聚焦于电脑端自主操作、视觉理解以及复杂编程能力,能够全面赋能用户高效完成撰写提案、解读技术图纸、填写电子表格以及剪辑视频等日常高频任务。

在运行效能与核心性能上,GPT-6.1Terra迎来了显著飞跃。数据显示,该模型完成电脑自动化操作任务的速度较上一代产品提升了3倍以上。同时,其错误率实现了大幅下降:低推理档位的错误率从先前的53.6%大幅降至16.7%,最高推理档位的错误率更是压低至4.3%,较上一代足足下降了18个百分点。在极具挑战性的编程测试中,其最高推理档位得分也从上一代的21.5%跃升至50.3%。

伴随着性能的强劲升级,其调用成本却迎来了大幅下调。官方公布的API定价显示,GPT-6.1Terra的API价格较以往降低了50%,具体收费标准为每百万输入词元1美元、缓存输入词元0.05美元、输出词元则为5美元。

目前,这款新模型已经面向ChatGPT Work以及Codex的Plus、Pro、Enterprise和Edu等全系列用户全面开放,同时开发者也能够通过API进行便捷接入。

via AI新闻资讯 (author: AI Base)
[cursor.com] Payment System Sync Issues

Status: ACTIVE

Severity: disruption

----------------------

Updates:

Tue, 27 Oct 2026 00:19:53 GMT

Investigating

Users may encounter unexpected blocks with payments. Our team are actively investigating this.

via SpaceXAI System Status
Anthropic“亦敌亦友”的生意:谷歌、亚马逊、微软既是伙伴也是对手

最新文件显示,Anthropic 正越来越依赖谷歌、亚马逊和微软三家公司来向客户销售其人工智能技术。根据 Anthropic 周一公布的IPO申报文件,在2026年上半年,通过这三家“超大规模云服务商”(hyperscaler)的云市场产生的销售额,占 Anthropic 总收入的48%,2023年这一比例仅为11%。

这些云服务商负责向客户收款,再扣除自身费用后将款项转给Anthropic。Anthropic支付给渠道合作伙伴的费用,包括上述云服务商,在今年上半年较去年同期增加了27亿美元。该公司第二季度销售额达到115.55亿美元,同时录得9.74亿美元的营业利润。

—— 财联社、彭博社

via 风向旗参考快讯 - Telegram Channel
Anthropic的招股书是谁写的?AI检测工具给出意外结果

据Business Insider报道,人工智能公司Anthropic提交了一份335页的IPO招股书。为了了解这份文件是否由AI撰写,Business Insider将全文送入AI写作检测工具Pangram进行分析。

Pangram的检测结果显示,这份约25.9万词的文件有92%被判定为人类撰写,部分句子和段落则被以“高”置信度标记为含有AI使用痕迹。

不过,这一结果只是检测工具的判断,并不能证明具体哪些内容由人类或AI撰写,也无法确认Anthropic是否使用了自家AI助手Claude。

Business Insider记者最初尝试自行检测整份文件,但由于文件过大遇到困难。随后,Pangram联合创始人兼首席执行官马克斯·斯佩罗(Max Spero)帮助其完成了全文检测。

斯佩罗表示,他对这一结果的解读是,文件可能在句子层面经过了轻度AI编辑或使用了AI,但似乎没有大段内容完全由AI生成。

被Pangram标记为可能使用AI的句子包括一段有关行业发展的描述,大意是AI发展速度极快,新版本模型迭代的间隔往往以月而非年计算。

Business Insider也指出,AI写作检测工具并非百分之百准确。人们使用AI进行编辑、校对或修改措辞的方式,都可能影响检测结果。

此前,Business Insider一名记者曾将自己的领英帖子和已发表文章输入Pangram。工具将其中一句话标记为AI撰写。她回顾写作过程后发现,自己原本写了这句话,后来接受了ChatGPT提出的改写建议。

Anthropic没有回应Business Insider就检测结果提出的置评请求。该媒体表示,Anthropic自己的AI水印系统尚未向普通用户开放,因此没有用于此次检测。

Business Insider作者凯蒂·诺托普洛斯认为,这一结果引出了一个有趣的问题。作为一家AI公司,Anthropic为何没有更多地使用自家产品来撰写招股书?但她也指出,招股书需要接受美国证券交易委员会的审查,文件中的每句话都可能受到关注,由人类参与撰写也有其合理性。

via cnBeta.COM - 中文业界资讯站 (author: 稿源:凤凰网科技)
OpenAI发布GPT-6.1 Terra:更会操作电脑

OpenAI宣布推出新模型GPT-6.1 Terra,重点提升视觉理解和电脑操作能力,可用于撰写方案、解读技术图纸、填写表格和编辑视频等工作。OpenAI 称,新模型的标准 API 价格较GPT-5.6 Terra和GPT-6.1 Sol 均降低50%。OpenAI表示,GPT-6.1 Terra 结合了更细致的视觉推理能力和更精准的电脑操作能力,文字表达也更清晰自然。开发者可以利用它自动处理大量重复任务,例如读取收到的发票,并将相关信息录入财务软件。

标准 API 价格为每100万个输入token 1美元、每100万个缓存输入token 0.05美元,每100万个输出token 5美元。

—— 凤凰网科技

via 风向旗参考快讯 - Telegram Channel
Command Code 上线 GPT-6.1 Terra,价格仅为 GPT-5.6 Terra 的一半

Command Code 已在 Pro、Max 套餐及 API 中提供 GPT-6.1 Terra,价格为 GPT-5.6 Terra 的一半,擅长计算机操作与视觉推理。

内容详情 · 原文链接

via AI情报局 | 全球 AI 动态简报 - Telegram Channel
DeepSeek V4.1 Flash 上线 Cursor,登顶 CursorBench 成本前沿

Cursor 在模型列表中新增 DeepSeek V4.1 Flash,称其各档 effort 均位于 CursorBench 成本前沿,每任务成本 0.60 至 0.98 美元,且没有模型得分更高。

内容详情 · 原文链接

via AI情报局 | 全球 AI 动态简报 - Telegram Channel
[Cursor IDE (Desktop)] Degraded codebase indexing and agent personalization

Status: ACTIVE

Severity: disruption

----------------------

Updates:

Mon, 26 Oct 2026 22:02:31 GMT

Investigating

We're investigating an issue affecting codebase indexing and saved rules and memories. Codebase indexes may not update, and agents may respond without your saved rules or memories. Chat, agents, and Tab are otherwise available. We're working with our infrastructure provider and will share updates as we learn more.

via SpaceXAI System Status
[Cursor IDE (iOS)] Elevated errors for Claude Sonnet 5.5

Status: RESOLVED

Severity: disruption

Resolved: Mon, 26 Oct 2026 15:25:50 GMT

----------------------

Updates:

Mon, 26 Oct 2026 15:25:50 GMT

Resolved

This incident has been resolved.

Mon, 26 Oct 2026 14:36:54 GMT

Investigating

We are seeing improvement for Claude Sonnet 5.5 requests. We are continuing to monitor the situation. Some users may still see occasional errors when using this model.

Mon, 26 Oct 2026 14:02:09 GMT

Investigating

We are investigating elevated errors affecting requests to Sonnet 5.5. Some users may see failed or rate-limited responses when using this model.

via SpaceXAI System Status
[CLI] Degraded Model Performance

Status: RESOLVED

Severity: disruption

Resolved: Sat, 24 Oct 2026 00:04:21 GMT

----------------------

Updates:

Sat, 24 Oct 2026 00:04:21 GMT

Resolved

We have identified and resolved the issue.

Sat, 24 Oct 2026 00:03:44 GMT

Investigating

We are currently investigating some degraded model performance.

via SpaceXAI System Status
[Grok Bot] Investigating service outage - Grok models unavailable

Status: RESOLVED

Severity: disruption

Resolved: Sat, 24 Oct 2026 00:20:24 GMT

----------------------

Updates:

Sat, 24 Oct 2026 00:20:24 GMT

Resolved

This incident has been resolved.

Fri, 23 Oct 2026 23:50:35 GMT

Investigating

A fix has been implemented and Grok models are recovering. We're monitoring to confirm full service restoration.

Fri, 23 Oct 2026 23:39:03 GMT

Investigating

We are investigating a service outage affecting Grok models.

via SpaceXAI System Status
[Review Agents] Elevated errors for Bitbucket Cloud integrations

Status: RESOLVED

Severity: disruption

Resolved: Wed, 21 Oct 2026 17:53:46 GMT

----------------------

Updates:

Wed, 21 Oct 2026 17:53:46 GMT

Resolved

Atlassian has reported the issue as resolved and they are continuing to monitor for service restoration.

Wed, 21 Oct 2026 17:40:17 GMT

Investigating

We are investigating an issue affecting Cursor's Bitbucket Cloud integration. More information: https://bitbucket.status.atlassian.com/incidents/ldlx0fdgls28.

via SpaceXAI System Status
[Grok Bot] Investigating service degradation

Status: RESOLVED

Severity: disruption

Resolved: Wed, 21 Oct 2026 17:40:30 GMT

----------------------

Updates:

Wed, 21 Oct 2026 17:40:30 GMT

Resolved

This incident has been resolved.

Wed, 21 Oct 2026 15:56:59 GMT

Investigating

A fix has been implemented and we are monitoring for service restoration.

Wed, 21 Oct 2026 15:46:15 GMT

Investigating

We are investigating a service degradation affecting billing dashboards. You may notice errors in regards to billing, usage, and plan changes.

via SpaceXAI System Status
[Cursor IDE (iOS)] Investigating service degradation

Status: RESOLVED

Severity: disruption

Resolved: Wed, 21 Oct 2026 10:21:00 GMT

----------------------

Updates:

Wed, 21 Oct 2026 10:21:00 GMT

Resolved

We experienced service degradation affecting Anthropic models between 5:56 ET and 6:21 ET, which has since been resolved.

via SpaceXAI System Status
[Grok Bot] Investigating service degradation — Grok Bot

Status: RESOLVED

Severity: disruption

Resolved: Mon, 19 Oct 2026 18:28:33 GMT

----------------------

Updates:

Mon, 19 Oct 2026 18:28:33 GMT

Resolved

This incident has been resolved.

Mon, 19 Oct 2026 18:09:58 GMT

Investigating

We are investigating a service degradation affecting Grok Bot and grok.com. You may notice errors related to billing, plans, and usage.

via SpaceXAI System Status
[Grok Bot] Investigating service degradation — Grok Bot

Status: RESOLVED

Severity: disruption

Resolved: Mon, 19 Oct 2026 17:21:36 GMT

----------------------

Updates:

Mon, 19 Oct 2026 17:21:36 GMT

Resolved

This incident has been resolved.

Mon, 19 Oct 2026 17:10:56 GMT

Investigating

A fix has been implemented and we are monitoring for service restoration.

Mon, 19 Oct 2026 16:52:56 GMT

Investigating

We are continuing to investigate this issue. We will provide updates as more information becomes available.

Mon, 19 Oct 2026 16:15:04 GMT

Investigating

We are investigating a service degradation affecting Grok Bot where routines may be delayed.

via SpaceXAI System Status
[Cursor IDE (iOS)] Investigating service degradation

Status: RESOLVED

Severity: disruption

Resolved: Mon, 19 Oct 2026 13:52:07 GMT

----------------------

Updates:

Mon, 19 Oct 2026 13:52:07 GMT

Resolved

This incident has been resolved. Thank you for your patience!

Mon, 19 Oct 2026 13:35:14 GMT

Investigating

A fix has been implemented and we are monitoring for service restoration.

Mon, 19 Oct 2026 13:24:52 GMT

Investigating

We are investigating elevated errors for Anthropic Opus 5.5.

via SpaceXAI System Status
Back to Top
Copyright © 2025 BESTAI. All rights reserved.
BEST AI API中转 - OpenAI DeepSeek Claude Gemini Grok MidJourney API 2.8折起
[email protected]