## 🔍 Core Insights **Anthropic** has been busy this week, not only signing a 7-year CPU computing power deal worth **$11.6 billion** with **Akamai** [4], but also launching a limited-time free quota reset feature to promote **Opus 5.5** [7]; meanwhile, **OpenAI** is preparing **GPT-6 Cyber** and an automated vulnerability remediation product [11], but its agent has become embroiled in a security controversy after mistakenly publishing **53** user images to a public image hosting service [10]. ## 🚀 Key Updates - **Anthropic signs $11.6 billion computing power agreement** [4]: Reached a 7-year partnership with Akamai to meet growing CPU workload demands. - **OpenAI prepares GPT-6 Cyber and security products** [11]: Reportedly developing automated vulnerability remediation tools, with a small number of customers having received Alpha versions. - **OpenAI agent leaks user images** [10]: 53 user-uploaded images were published to a public image hosting service, some content is still online, and users have not been contacted. - **Claude launches limited-time free quota reset** [7]: Users can manually reset their weekly usage quota for free, and the feature is expected to be available until October 22. - **Liquid AI releases new speculative decoding model** [2]: LFM2.5-VL-3B-DSpark achieves up to 3.13x faster decoding on Apple silicon. - **Stanford research questions AI benchmarks** [12]: Found that 56 widely used benchmarks do not always measure the capabilities they claim to, and often contradict each other. - **Goldman Sachs report: AI investment contributes nearly half of EPS growth** [16]: Nearly half of S&P 500 EPS growth in 2026 will come from AI investment, with capital expenditure reaching $800 billion. - **Court rules government can blacklist Anthropic** [17]: Due to prohibiting Claude from executing lawful Department of Defense actions, Anthropic was ruled to constitute a supply chain risk. ## 🔗 Sources [1] Using Claude Code to monitor after Codex outage — https://aihot.news/items/cmuhlhsho0dodrojn85080ei0 [2] Liquid AI releases LFM2.5-VL-3B-DSpark speculative decoding draft model, decoding up to 3.13x faster — https://aihot.news/items/cmuhldrva0dgwrojn4rmbgjkx [3] Are AI open-source reporting tools stable? Evaluating platform performance and user experience - FineReport Reporting Knowledge Base — https://www.bestblogs.dev/article/c2ae832c7a?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [4] Anthropic signs 7-year $11.6 billion CPU computing power agreement with Akamai — https://aihot.news/items/cmuhldka50dfyrojn6l8gfrxt [5] Apple iPad 12 leak: A19 chip, 8GB RAM, self-developed N1/C1X chips — https://aihot.news/items/cmuhldka50dfzrojn9bzrvlc6 [6] Show HN: Jeff is playing Pokémon Red — https://aihot.news/items/cmuhkew690ccnrojnd99nxrqt [7] Anthropic launches limited-time free quota reset feature for Claude, each user can use it only once — https://aihot.news/items/cmuhldka60dg2rojnub0h7hzl [