{"id":1438,"date":"2026-07-11T09:16:13","date_gmt":"2026-07-11T01:16:13","guid":{"rendered":"https:\/\/blog.liu-qi.cn\/2026\/07\/11\/x-daily-2026-07-10\/"},"modified":"2026-07-11T09:16:13","modified_gmt":"2026-07-11T01:16:13","slug":"x-daily-2026-07-10","status":"publish","type":"post","link":"https:\/\/en.blog.liu-qi.cn\/2026\/07\/11\/x-daily-2026-07-10\/","title":{"rendered":"X Platform July 10 AI Brief | GPT-5.6 Officially Released, Zuckerberg Returns to X Platform for Counter-Launch, MiniMax Secures Massive Funding"},"content":{"rendered":"<p><strong>Key Highlights: GPT-5.6 officially released, Codex integrated into ChatGPT, Meta Muse Spark 1.1 launches as counter-move, Grok 4.5 leads multi-dimensional evaluations, MiniMax completes $2B new funding round, Tencent reportedly leads buyback of Manus, Ollama secures $65M Series B, OpenAI App CEO resigns.<\/strong><\/p>\n<h2 id=\"topic-3f60ffbea6\">GPT-5.6 Officially Released, Codex Integrated into ChatGPT, Product Lineup Fully Restructured<\/h2>\n<p>On July 9 local time, OpenAI held a live launch event, officially introducing the GPT-5.6 series of models. It also announced the merger of Codex with the ChatGPT desktop client and launched ChatGPT Work Agent and ChatGPT Sites hosting services. GPT-5.6 includes three tiers: Sol (flagship, excels at programming and complex Agent tasks), Terra (value-oriented, performance close to GPT-5.5 but at lower cost), and Luna (fastest and cheapest high-throughput model). OpenAI&#8217;s product lead @thsottiaux stated that GPT-5.6 Sol is the company&#8217;s best model for nearly all tasks, showing significant gaps compared to GPT-5.5 in programming, complex context, and strategic tasks. Sol adds multiple reasoning tiers from Medium to Ultra, with Ultra mode capable of invoking multiple sub-agents for parallel processing. Sam Altman revealed that GPT-5.6 has become the preferred model for Microsoft 365 Copilot. CEO Altman called it the &#8220;best blog post ever.&#8221;<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@thsottiaux: <a href=\"https:\/\/x.com\/thsottiaux\/status\/2075283734136525255\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/thsottiaux\/status\/2075283734136525255<\/a><\/li>\n<li>@OpenAIDevs: <a href=\"https:\/\/x.com\/OpenAIDevs\/status\/2075286157186003348\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/OpenAIDevs\/status\/2075286157186003348<\/a><\/li>\n<li>@OpenAIDevs: <a href=\"https:\/\/x.com\/OpenAIDevs\/status\/2075273992609599834\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/OpenAIDevs\/status\/2075273992609599834<\/a><\/li>\n<li>@OpenAIDevs: <a href=\"https:\/\/x.com\/OpenAIDevs\/status\/2075274080740380829\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/OpenAIDevs\/status\/2075274080740380829<\/a><\/li>\n<li>@OpenAIDevs: <a href=\"https:\/\/x.com\/OpenAIDevs\/status\/2075274093327470923\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/OpenAIDevs\/status\/2075274093327470923<\/a><\/li>\n<li>@OpenAIDevs: <a href=\"https:\/\/x.com\/OpenAIDevs\/status\/2075292793417973889\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/OpenAIDevs\/status\/2075292793417973889<\/a><\/li>\n<li>@sama: <a href=\"https:\/\/x.com\/sama\/status\/2075585386441789605\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/sama\/status\/2075585386441789605<\/a><\/li>\n<li>@sama: <a href=\"https:\/\/x.com\/sama\/status\/2075266471316615436\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/sama\/status\/2075266471316615436<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-fc837c4e4f\">Meta Releases Muse Spark 1.1, Zuckerberg Returns to X After Three Years to Directly Counter GPT-5.6<\/h2>\n<p>Approximately 3 hours before the GPT-5.6 live stream began, Meta founder Mark Zuckerberg suddenly posted about Muse Spark 1.1 on X, marking his first tweet in nearly three years. The model is positioned as the cheapest cutting-edge Agentic model on the market, scoring 88.1 (highest among all models) on the MCP Atlas tool-use evaluation, supporting a 1 million token context window, capable of delegating tasks to sub-agents running in parallel, and made available to developers via the Meta Model API. Multiple bloggers noted its price of $1.25\/$4.25, highlighting its high cost-effectiveness. Elon Musk commented that a CEO announcing a product directly on X is more engaging to the public than a press release, and Zuckerberg&#8217;s tweet has already garnered over 12 million free impressions.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@finkd: <a href=\"https:\/\/x.com\/finkd\/status\/2075218444056707458\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/finkd\/status\/2075218444056707458<\/a><\/li>\n<li>@elonmusk: <a href=\"https:\/\/x.com\/elonmusk\/status\/2075403875410928124\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/elonmusk\/status\/2075403875410928124<\/a><\/li>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075258457964593467\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075258457964593467<\/a><\/li>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075257352459305125\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075257352459305125<\/a><\/li>\n<li>@LufzzLiz: <a href=\"https:\/\/x.com\/LufzzLiz\/status\/2075254978873311585\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/LufzzLiz\/status\/2075254978873311585<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-1cef594315\">Grok 4.5 Continues to Lead Multi-Dimensional Evaluations, Grok Build Ecosystem Expands<\/h2>\n<p>Grok 4.5 continues to receive positive evaluations from various sources after its release. Multiple bloggers conducted comparative tests: it ranked first on the AutomationBench-AA with 51%, significantly outperformed Opus 4.8 and GPT-5.5 on the professional work benchmark (Snorkel), and tied for first with GPT-5.6 Codex on the SWE-Atlas-QnA evaluation with a score of 84. In mathematical research, Grok 4.5 constructed an explicit counterexample for Poisson semigroup hypercontractivity. Composio&#8217;s evaluation called it the &#8220;most persistent Agent model.&#8221; On the application front, the Grok Build feature continues to iterate: it supports creating FPS games, allows viewing all running sub-agents in a TUI, and adds the Grok Imagine image generation tool. Elon Musk revealed that Grok will soon support a context window of over 1 million tokens.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@XFreeze: <a href=\"https:\/\/x.com\/XFreeze\/status\/2075487843691229539\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/XFreeze\/status\/2075487843691229539<\/a><\/li>\n<li>@kevinnbass: <a href=\"https:\/\/x.com\/kevinnbass\/status\/2075303746397962662\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/kevinnbass\/status\/2075303746397962662<\/a><\/li>\n<li>@PI010101: <a href=\"https:\/\/x.com\/PI010101\/status\/2075325611799953778\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/PI010101\/status\/2075325611799953778<\/a><\/li>\n<li>@composio: <a href=\"https:\/\/x.com\/composio\/status\/2075225191160889786\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/composio\/status\/2075225191160889786<\/a><\/li>\n<li>@tetsuoai: <a href=\"https:\/\/x.com\/tetsuoai\/status\/2075351242025398678\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/tetsuoai\/status\/2075351242025398678<\/a><\/li>\n<li>@elonmusk: <a href=\"https:\/\/x.com\/elonmusk\/status\/2075254074224623692\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/elonmusk\/status\/2075254074224623692<\/a><\/li>\n<li>@JasonBud: <a href=\"https:\/\/x.com\/JasonBud\/status\/2075259469844365375\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/JasonBud\/status\/2075259469844365375<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-2620b85d3b\">MiniMax Completes $2 Billion New Funding Round, Founder Vows Not to Take Salary Until AGI is Achieved<\/h2>\n<p>According to reposts by multiple bloggers, MiniMax announced the completion of a new $2 billion funding round. Concurrently, founder IO made three long-term commitments to the team: he will not take a personal salary until AGI is achieved; over the next four years, he will allocate shares equivalent to 4% of the company&#8217;s total equity from his personal holdings to reward employees; and an additional 1% will be dedicated to supporting the open-source community. One blogger commented that this is one of the most exciting moments in the history of global AI.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075423664229814611\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075423664229814611<\/a><\/li>\n<li>@imwsl90: <a href=\"https:\/\/x.com\/imwsl90\/status\/2075422297847210084\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/imwsl90\/status\/2075422297847210084<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-9696fd4b46\">Tencent Reportedly Leads $2 Billion Buyout to Repurchase Manus Equity from Meta<\/h2>\n<p>According to the Financial Times, Tencent is negotiating to become Manus&#8217;s largest shareholder, participating in a roughly $2 billion (approximately 13.5 billion RMB) buyback of the company from Meta. Participants include existing investors like ZhenFund and HSG, as well as management. Upon completion, Manus will operate independently under a Singapore entity and will not be integrated into Tencent&#8217;s system. The report states that Manus&#8217;s ARR has grown rapidly from $100 million to nearly $500 million (approximately 3.4 billion RMB) since its acquisition by Meta. Tencent is attracted by Manus&#8217;s Agent collaboration capabilities, and founder Xiao Hong has deep ties with Tencent. Some insiders caution that maintaining growth after leaving the Meta ecosystem remains uncertain.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075444454606193089\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075444454606193089<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-c8ffe11759\">OpenAI App CEO Fidji Simo Resigns Due to Health Reasons<\/h2>\n<p>OpenAI&#8217;s App business CEO, Fidji Simo, announced her resignation from her full-time role due to worsening chronic illness, transitioning to a part-time advisor position. Simo previously spent over a decade at Facebook and served as Instacart&#8217;s CEO for four years before joining OpenAI&#8217;s board in 2023. She was appointed App CEO in May 2025, leading the restructuring of the App business unit and implementing a product general manager system. Altman publicly expressed great sadness and wished her a speedy recovery. Greg Brockman has been overseeing product work during her medical leave.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075421692265193896\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075421692265193896<\/a><\/li>\n<li>@sama: <a href=\"https:\/\/x.com\/sama\/status\/2075354679031067058\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/sama\/status\/2075354679031067058<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-a089d7db5b\">Ollama Completes $65 Million Series B Funding, Shifts Towards Hybrid Inference<\/h2>\n<p>Ollama announced it has secured $65 million in Series B funding, led by Theory Ventures, with participation from Benchmark, 8VC, YC, and others. Ollama currently serves 8.9 million developers, has 176k stars on GitHub, and has a team of only 14 people. The company is transitioning to a &#8220;local + cloud&#8221; hybrid inference model, where small models run locally and large models connect to Ollama Cloud.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@ollama: <a href=\"https:\/\/x.com\/ollama\/status\/2075211168407503016\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/ollama\/status\/2075211168407503016<\/a><\/li>\n<li>@MaxForAI: <a href=\"https:\/\/x.com\/MaxForAI\/status\/2075288258901311724\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/MaxForAI\/status\/2075288258901311724<\/a><\/li>\n<\/ul>\n<h2 id=\"topic-93e6428124\">Other News: Cloudflare Drop, 1X Dexterous Hand, Claude Update<\/h2>\n<p>Cloudflare released the Drop feature: drag and drop a folder to deploy a static website on its global edge network, no registration required, and it can be tested and shared within about an hour. 1X (NEO) unveiled a dexterous robotic hand with 25 tendons, capable of screwing in light bulbs, assembling Legos, and zipping zippers, achieving fine manipulation. Claude added a Reflect function, which, like a screen time tool, provides 1-12 month AI usage reports, analyzing key topics, usage periods, and task types. Elon Musk publicly admitted he misjudged Anthropic, stating they are clearly the current AI leader.<\/p>\n<p>Sources:<\/p>\n<ul>\n<li>@xiaohu: <a href=\"https:\/\/x.com\/xiaohu\/status\/2075395846213091769\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/xiaohu\/status\/2075395846213091769<\/a><\/li>\n<li>@op7418: <a href=\"https:\/\/x.com\/op7418\/status\/2075375991980278010\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/op7418\/status\/2075375991980278010<\/a><\/li>\n<li>@xiaohu: <a href=\"https:\/\/x.com\/xiaohu\/status\/2075498708914057616\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/xiaohu\/status\/2075498708914057616<\/a><\/li>\n<li>@elonmusk: <a href=\"https:\/\/x.com\/elonmusk\/status\/2075278580955685036\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/x.com\/elonmusk\/status\/2075278580955685036<\/a><\/li>\n<\/ul>\n<p>Statistics:<\/p>\n<ul>\n<li>Timeline items scanned=480<\/li>\n<li>Bloggers matched=43<\/li>\n<li>Total tweets matched=338<\/li>\n<li>Weighted tweet score=264.35<\/li>\n<li>Original tweets=160<\/li>\n<li>RT tweets=77<\/li>\n<li>Crawl attempts=3<\/li>\n<li>Boundary coverage status=tail_confidently_crossed_target_boundary<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>OpenAI releases GPT-5.6 and restructures its product lineup, while Meta and Grok also roll out major updates, intensifying industry competition.<\/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-1438","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\/1438","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=1438"}],"version-history":[{"count":0,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/posts\/1438\/revisions"}],"wp:attachment":[{"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/media?parent=1438"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/categories?post=1438"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/en.blog.liu-qi.cn\/index.php\/wp-json\/wp\/v2\/tags?post=1438"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}