Multiple bloggers mention that Opus 4.6 outperforms Opus 4.7 in several benchmarks, with the user satisfaction “frustration index” (lower is better) ranking Opus 4.6 first. The evaluation also covers models like Opus 4.7, GPT 5.5, GPT 5.4, and Gemini 3.1.<\/p>\n
Sources:<\/p>\n
Some bloggers note that Grok’s connection feature now supports synchronization across Web, iOS, and Android, allowing users to authorize once on the web for multi-device sync.<\/p>\n
Sources:<\/p>\n
Multiple bloggers report that Anthropic has collaborated with SpaceX (xAI) on computing power, with SpaceX opening access to the Colossus 1 supercomputer (over 220,000 NVIDIA GPUs, 300+ megawatts) to Anthropic, significantly increasing usage limits for Claude Code and Claude API. Anthropic has also expressed interest in partnering with SpaceX to build “gigawatt-scale orbital AI computing power.”<\/p>\n
Sources:<\/p>\n
Some bloggers mention that court proceedings in Elon Musk’s lawsuit against OpenAI have revealed text message records from November 2023 between Altman and former CTO Mira Murati. Murati informed Altman that “they’ve decided, they just don’t want you,” and described the new CEO Emmett Shear as “an unknown Twitch guy.” Altman inquired about transferring OpenAI IP to Anthropic, to which Murati responded: “They just don’t want AGI to be in your hands.”<\/p>\n
Sources:<\/p>\n
Some bloggers report that OpenAI, along with AMD, Broadcom, Intel, Microsoft, and NVIDIA, has open-sourced a new network protocol called MRC (Multi-Path Reliable Connection). This protocol disperses data packets across hundreds of paths for concurrent transmission, bypassing traditional routing convergence bottlenecks, and has been deployed on OpenAI’s largest GB200 supercomputer, including the Stargate project.<\/p>\n
Sources:<\/p>\n
Some bloggers highlight a batch of AI development tools: Refero Styles can automatically extract website design styles and generate DESIGN.md, supporting documentation for numerous quality products like Linear and Stripe; Open Slide is an open-source project for creating presentations with React, integrating over 1500 brand logos; Flue is a TypeScript Agent development framework similar to Claude Code; the Codex desktop version is now available for download, supporting third-party APIs and proxy stations.<\/p>\n
Sources:<\/p>\n
Some bloggers mention that Anthropic announced at the Code with Claude event: the Dreaming function in Managed Agents enters research preview, while Multiagent orchestration and Webhook are officially open for public beta.<\/p>\n
Sources:<\/p>\n
Some bloggers report that Cursor version 3.3 introduces visual analysis of Agent context usage, categorized by Rules, Skills, MCPs, and Subagents, helping developers diagnose context overflow issues.<\/p>\n
Sources:<\/p>\n
Some bloggers mention that TRAE SOLO, under ByteDance, has officially fully opened its mobile version, enabling synchronization across mobile, Web, and desktop. The mobile device is not used as a remote desktop but rather for assigning tasks to Agents, voice discussions, and dispatching work on the go, with the mobile end linked to cloud environments and multiple local devices. The Agent’s timed Loop mechanism (repeating every minute\/five minutes\/daily) further popularizes automating repetitive tasks.<\/p>\n
Sources:<\/p>\n
Some bloggers report that a set of programmer-themed clay-style 3D icons has been open-sourced, including figures like the Debug dinosaur, Docker whale, Git Merge octopus, and AI Agent robot, generated with consistent style through unified master prompts. Additionally, a “Million-Follower Challenge Plan Kanban” created with Codex in one hour via Vibe Coding is now open-sourced, supporting one-click free deployment on Vercel.<\/p>\n
Sources:<\/p>\n
Yesterday’s scraping statistics (source: manifest-2026-05-07.json)<\/p>\n
Unexpected results emerge in AI model evaluations, while major collaborations on computing power and open-source agreements draw industry attention.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[19],"class_list":["post-1179","post","type-post","status-publish","format-standard","hentry","category-brief","tag-x--ai-"],"_links":{"self":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1179","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/comments?post=1179"}],"version-history":[{"count":0,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1179\/revisions"}],"wp:attachment":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/media?parent=1179"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/categories?post=1179"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/tags?post=1179"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}