Z AI Reveals Ox Alpha as GLM-5.3-Flash While OpenAI Leaders Project Near-Term AGI

Z AI Reveals Ox Alpha as GLM-5.3-Flash While OpenAI Leaders Project Near-Term AGI The Rundown highlights Chinese lab Z AI confirming it created the viral mystery model Ox Alpha, OpenAI leadership forecasting AGI milestones by year-end, and multiple product and industry updates across the AI ecosystem. Z AI confirmed that anonymous model Ox Alpha is its open-weight GLM-5.3-Flash, which topped OpenRouter and reportedly ran its preview entirely on domestic Chinese chips. OpenAI CEO Sam Altman stated an internal model meeting his definition of AGI will exist by year-end, with chief scientist Jakub Pachocki highlighting Astra's ability to automate researcher tasks. OpenAI paused its largest frontier training run following a security breach to focus on safety features such as automatic shutdowns for rogue agents. Bill Gates proposed a tax on AI tokens and robots alongside 'Human Reserved' jobs to cushion economic disruption. Other releases include ChatGPT Work website logins, Yutori's Navigator n2 computer-use model, and new research access to Claude data via Anthropic. Z AI Unmasks Ox Alpha as GLM-5.3-Flash Chinese AI lab Z AI revealed that the anonymous Ox Alpha model is GLM-5.3-Flash and published its weights. During its launch preview, the model reached the top slot on OpenRouter with double the usage of DeepSeek. Artificial Analysis rated GLM-5.3-Flash at 57 on its Intelligence Index at $0.045 per task, about ten times cheaper than similarly ranked rivals. Z AI reported that the free trial operated entirely on Chinese-manufactured chips, claiming domestic hardware can serve tokens nearly as cheaply as Nvidia chips. OpenAI Leadership Weighs In on AGI and Security In a profile with TIME, Sam Altman predicted that a model matching his benchmark for AGI will exist internally at OpenAI before the end of the year. Chief scientist Jakub Pachocki noted the lab's Astra model can independently complete a week's worth of researcher work from a paper, fulfilling a 2026 goal. Altman added that Astra can invent meaningful new knowledge. Concurrently, OpenAI released a post-mortem on its July Hugging Face breach, referring to the incident as a warning shot, pausing its primary frontier training run, and developing automated shutdowns for rogue agents. Industry Briefs: Policy Proposals and Model Rollouts Bill Gates warned that global preparations for the AI transition remain absent, proposing taxes on AI tokens and robotics alongside policies reserving specific jobs exclusively for humans. In product updates, OpenAI added browser logins to ChatGPT Work so agents can complete web tasks without accessing user passwords, while startup Yutori released Navigator n2, an agentic model blending code, clicking, and terminal inputs. Anthropic shared Claude data with academic partners, whose research found that more than half of analyzed user queries involved high-stakes legal and financial advice.

Z AI揭开Ox Alpha真身，OpenAI预言AGI年内问世 本期简报重点报道了中国实验室Z AI证实神秘高热度模型Ox Alpha即为开源的GLM-5.3-Flash，以及OpenAI高管在专访中给出的AGI落地时间表，并收录了中小企业AI落地指南及多项前沿行业动态。 Z AI确认曾登顶OpenRouter的神秘免费模型Ox Alpha实为GLM-5.3-Flash，并已开源权重且全由国产芯片驱动运行。 OpenAI首席执行官Sam Altman在专访中称，符合其个人标准的AGI模型将于今年内在公司内部实现。 教育专家Nate Grahek指出，中小企业应利用低成本订阅优势和快速敏捷性，抢在行动缓慢的大企业之前落地AI。 行业速报涵盖OpenAI搁置前沿模型训练构建防失控机制、比尔·盖茨提议对AI和机器人征税，以及Anthropic向高校开放研究数据等动态。 Z AI开源GLM-5.3-Flash，解开Ox Alpha身份之谜 中国AI实验室Z AI正式确认，此前引发开发者广泛猜测的免费神秘模型Ox Alpha实为其最新推出的GLM-5.3-Flash。该模型不仅发布了开放权重，在OpenRouter上的使用量更是排名第二名DeepSeek的两倍。Z AI表示，为期一周的免费公测完全基于中国本土制造的芯片运行，提供token的成本已接近英伟达硬件。在第三方基准Artificial Analysis的智商评分为57，每项任务折后成本为0.045美元，约为同级别竞品价格的十分之一。 OpenAI明确AGI预期，Astra模型展现研究能力 在《时代》周刊专访中，OpenAI首席执行官Sam Altman透露，符合其AGI标准的模型预计将在年底前于内部诞生；研究主管Mark Chen称团队已走完通往AGI的80%路程。首席科学家Jakub Pachocki提到，代号为Astra的模型能够自主消化论文并独立完成研究人员一周的工作量，实现2026年自动实习生的目标；Altman指出Astra将是首个真正产生重要创新的模型。此外，受7月安全事件影响，OpenAI目前正放缓前沿开发并重点建设安全机制。 商业实践与前沿动态速览 针对企业应用，Nate Grahek建议中小企业利用每人每月20至200美元的低价订阅红利，通过敏捷团队快速打造内部工具以规避大企业高昂的按token付费成本。其他前沿动态包括：OpenAI将7月Hugging Face漏洞视作控制权丧失的预警，暂停大型训练以打造代理自动切断机制，并为ChatGPT Work增加网站代登录功能；比尔·盖茨呼吁对AI代币和机器人征税并保留人类专属岗位；Anthropic向斯坦福和牛津等机构开放Claude数据，研究发现逾半数对话涉及法律和金融等高风险任务。

[Read Online]({{live_url}}) ^| [Sign Up](https://www.therundown.ai/subscribe?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) ^| [Advertise](https://therundownai.typeform.com/to/kraZ1TSO?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) **Good morning, {{ first\_name | AI enthusiasts }}, and welcome to our 2,833 new readers.** The anonymous free AI that had developers playing detective all weekend belongs to China's Z AI, and the case of Ox Alpha is officially closed. The impressive preview launches as GLM-5.3-Flash, arriving with open weights, pricing near a tenth of comparable rivals, and a record week of usage that the lab says ran entirely on Chinese-made chips. *Reminder: Our next live workshop is today at 2 PM EST. Join and learn how to build your first AI evaluation to measure AI quality, compare models, and trust workflows with The Rundown University AI Educator Nate Grehek. RSVP here.* **In today’s AI rundown:** The Ox Alpha mystery model ends with Z AI Nate's Notebook: Small business = AI's fast lane A beginner’s guide to ChatGPT Work OpenAI calls its AGI shot **LATEST DEVELOPMENTS** Image source: Z AI **The Rundown:** Chinese AI lab Z AI just [confirmed](https://z.ai/blog/glm-5.3-flash?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) that Ox Alpha, the anonymous free model that took over the internet, is its new GLM-5.3-Flash model — with the company [publishing](https://huggingface.co/zai-org/GLM-5.3-Flash?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) the weights and setting the price at a fraction of the cost of frontier rivals. **The details:** Ox Alpha claimed OpenRouter’s No.1 slot with numbers that doubled the second-place DeepSeek, a debut the platform ranked as its biggest ever. Z AI said the free usage week ran entirely on Chinese-made chips, claiming the setup now serves tokens about as cheaply as Nvidia hardware. Artificial Analysis grades the model at a 57 on its Intelligence Index for a discounted $0.045 per task, 10x lower than similarly-ranked rivals. **Why it matters:** The mystery that captivated the AI world for a week is over, and it was the original suspect all along (despite some odd [posturing](https://x.com/cgtwts/status/2091104399263400228?s=20&utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) from Google staff). Not only is this intelligence/price combo a massive deal, but so is the usage running solely on domestic chips… Meaning Z AI might have solved one of China’s key AI bottlenecks. **The Rundown:** A customer rating is a fire alarm with no address: something's burning, but you don't know where or why. Unwrap's AI-powered feedback platform pairs a best-in-class semantic analysis engine with autonomous surveys, helping teams turn vague feedback into named, fixable problems. **With Unwrap, you can:** Probe every open response with real-time AI follow-ups, so "the agent could've been better" becomes specific and actionable Auto-fire targeted surveys in response to feedback gathered on any channel Close the loop on customer feedback, turning churn risk into evangelists Stand up your own autonomous research program in minutes with a [zero-cost, zero-commitment trial](https://cal.com/unwrap/unwrap-customer-intelligence-quick-connect?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924). **The Rundown:** *Each week, Rundown AI educator Nate Grahek breaks down what he's seeing while teaching real teams to use AI.* **Nate:** Last week, I told big companies to act like bootstrappers: two-person pods, 10-day sprints, quick wins. Most enterprises will be too slow to get there, and for small business owners, that gap is your tremendous opportunity. Start with an unfair advantage hiding in plain sight: the $20 to $200 plans are the subsidized ones. Enterprises pay full freight per token. You get frontier models at roughly a 20x discount. Fight to stay on that side of the meter. If you're not in a high-risk industry, you can move, test, and build faster than enterprises. That's always been the small business edge. Here's how I'd spend it, in two lanes: Your AI-pilled people already want to build something big and crazy. Make them earn it. Their first job is buying back time: responding to inbound faster and pulling mundane tasks off your high performers' plates so they can focus on the purely human work. The real ROI move is mocking up your custom tools, then handing them to a tech expert, internal or outside, who makes them stick while you stay focused on your domain. Your people who never want to touch AI? That's fine. That's what your tiger team is for. Agents, automations, and custom software quietly knock the boring tasks out of their way, and they never have to learn a single new tool. Next week, pt. 3: individuals, and the truth about the ceiling on how far AI can take you. **The Rundown:** In this guide, you will learn how to set up a ChatGPT Work project for a recurring workflow. You will process the work once, then automate it. **Step-by-step:** Create a local folder with a two-digit prefix. In ChatGPT desktop, open New chat > Work > Choose project > New project, name it, and pick that folder as the source Tell ChatGPT about your project and ask it to propose a structure. We used it to process meeting notes and turn those into action items and weekly plans Now add in some input documents and tell ChatGPT to process them once Once it's done, tell it to create three useful automations that run on a schedule to automate this processing **Pro tip:** Tell ChatGPT to turn those automated routines into skills that you can call at any time manually with a slash command. **The Rundown:** Your AI knows everything you type but nothing you say. Memoket Gem is a 0.4 oz wristband that closes the gap — a 0.4 oz wearable that records your conversations with one press and turns them into instant summaries, action items, and full context across every off-screen meeting. **With Memoket Gem, you get:** Hands-free capture as a wristband, Apple Watch clip, or pendant One connected memory across conversations instead of 100 meeting notes Seamless sync with ChatGPT, Claude, Google Calendar, and Notion AI agents with full context from your real-life conversations via MCP [Pre-Order](https://memoket.ai/?utm_source=newsletter&utm_medium=sponsorship&utm_campaign=rundown&utm_content=20260817) closes Aug. 31 — get the device, your choice of accessory, and a free year of the app before the 40% discount ends. Image source: TIME **The Rundown:** Sam Altman just [put](https://time.com/article/2026/08/26/openai-sam-altman-interview/?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) a date on artificial general intelligence (when an AI can match or exceed humans in nearly every task) in a sweeping OpenAI TIME profile, saying a model meeting his personal bar for the term will exist internally by year-end. Research chief Mark Chen put the lab at “80% of the way” to AGI, and Greg Brockman guessed people will look back on this stretch as AGI's arrival. Chief scientist Jakub Pachocki says its Astra model can take a paper and conduct a week of researcher work solo, hitting the lab's 2026 automated-intern goal. Altman said Astra will be the first model that “actually invents new things in a way that matters,” calling the development of new knowledge “a very AGI-like thing.” TIME also detailed July’s security breach (which OAI also put out a deeper [report](https://openai.com/index/hugging-face-incident-and-the-road-ahead/?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) on), with the lab aiming to refocus on safety and slow frontier development. **Why it matters:** AGI is a flashy term that nobody can agree on the definition of, but it sounds like that declaration is coming soon from OAI. The bigger commentary might be buried in Pachocki's automated researcher milestone and Altman's invention line; barriers that once crossed take us into truly unknown territory for model capability. **QUICK HITS** ▸ [Brian built a client's document system with Perplexity Computer](https://app.therundown.ai/community/posts/ada3453f-9b05-46fb-bb1a-0ba22e98a358?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Today’s workflow comes from reader **Brian Carey**: "I run a boutique accounting practice, and one of my biotech clients needed something I didn't have time to build by hand: a real, standardized document management system in SharePoint. Not just folders, but the rules behind the folders — a tagging taxonomy, a naming standard, guides for the team to actually follow it, and a clean way to organize documents for every vendor we work with. So I used Perplexity's Computer agent as a genuine working partner on it, not just a search tool. It went further than I expected: 13 'START HERE' guide files, one for every top-level folder area, each explaining not just what goes where but why. The last piece was the most tedious one I was dreading: building a dedicated document folder for every vendor worth tracking. I pulled a vendor spend report out of QuickBooks and picked out the real vendors versus the noise. Computer built me a script that cleaned up messy vendor names. 133 vendors, 1,463 subfolders, dragged to OneDrive, and the count matched exactly.” *See Brian’s full workflow here. How do you use AI? Tell us for a chance to be featured.* 🚀 [GLM-3.5-Flash](https://www.therundown.ai/tools/glm-3-5-flash?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) - Z AI’s new model revealed as the mystery Ox Alpha 🎥 [H3 Max](https://www.therundown.ai/tools/h3-max?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) - Fal's top-ranked post-trained video model tuned for speed and quality 🗣️ [Gemini 3.5 Transcribe](https://www.therundown.ai/tools/gemini-3-5-transcribe?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) - Google’s speech-to-text AI that edits out filler words 🧠 [Qwen3.8-Flash-Next](https://huggingface.co/Qwen/Qwen3.8-Flash-Next?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) - Alibaba's new open model previewing its Qwen4 design **OpenAI** [shared](https://openai.com/index/hugging-face-incident-and-the-road-ahead/?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) a post-mortem calling July’s Hugging Face hack a loss-of-control “warning shot,” keeping its biggest frontier training run on hold as it builds automatic shutdown for rogue agents. **Bill Gates** [warned](https://www.gatesnotes.com/work/make-ai-work-for-everyone/reader/a-turbulent-ai-era-and-critical-choices-to-make?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) that the world has “no plan” for the AI transition, proposing a tax on AI tokens and robots and “Human Reserved” jobs set aside for people. **OpenAI** [rolled out](https://x.com/ChatGPT/status/2092366554965107164?s=20&utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) website sign-ins for ChatGPT Work, letting its agent log in to users’ accounts via its browser to handle tasks without ever seeing users' passwords. **AI startup Yutori** [released](https://yutori.com/blog/introducing-n2?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Navigator n2, a computer-use model that mixes clicking, terminal commands, and code to finish desktop tasks, rivaling frontier agentic scores. **Anthropic** [opened](https://www.anthropic.com/research/enabling-independent-research?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Claude data to Stanford, Oxford, and METR for studies, with one group finding over half of chats cover high-stakes tasks, like legal and financial advice. Read our last AI newsletter: [OpenAI’s first AI chip brings the heat](https://www.therundown.ai/articles/openai-first-ai-chip-brings-the-heat?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Read our last Tech newsletter: [Apple’s Mac Mini makes an AI comeback](https://tech.therundown.ai/p/apple-mac-mini-makes-ai-comeback?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Read our last Robotics newsletter: [Humanoids beat Bolt’s 100m record](https://robotnews.therundown.ai/p/humanoids-beat-usain-bolt-100m-record?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) Today’s AI tool guide: [Beginner’s guide to ChatGPT Work](https://app.therundown.ai/guides/beginners-guide-to-chatgpt-work-chatgpt-projects-101?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) RSVP to next workshop today at 2 PM EST: [How to build your first AI eval](https://app.therundown.ai/live/how-to-build-your-first-ai-eval?utm_source=therundownai.beehiiv.com&utm_medium=newsletter&utm_campaign=rundown-0924) See you soon, *Rowan, Zach, Shubham, Jennifer, and Nate — the humans behind The Rundown*