OpenAI launches GPT-6 Astra, claims AGI era, scores 99.9% on ARC-AGI-3
OpenAI released GPT-6 Astra, its next-generation model focused on computer and browser use, calling it the world's most intelligent and aligned model. The company claims it marks the start of the AGI era, with president Greg Brockman stating 'Welcome to the AGI era.' Astra scored 99.9% on ARC-AGI-3 with a provider adapter harness, surpassing human baseline in action efficiency, and was trained on over 100,000 GPUs at the Stargate site in Texas.
Coverage timeline
TechCrunch AILucas Ropek
OpenAI claims that Astra represents "a new frontier on computer and browser use," and that it handles tasks with unmatched "speed, accuracy, and safety."
Hacker Newsmaskil
https://thenewstack.io/openai-gpt6-astra-benchmarks/ , image: https://cdn.thenewstack.io/media/2026/09/358eb84a-screenshot... https://venturebeat.com/technology/welcome-to-the-agi-era-op... https://www.theverge.com/ai-artificial-intelligence/988334/o... https://twitter.com/OpenAI/status/2095527557924082061
Techmeme
Maxwell Zeff / Wired : OpenAI calls GPT-6 Astra the “world's best computer use model”; in tests, it booked DMV appointments and searched job listings faster than the average person — OpenAI leaders think the company's next generation model, which excels at computer use and coding, may mark a major milestone in AI development.
Ina Fried / Axios : OpenAI says Astra was built on its largest-ever training run, using more than 100,000 GPUs at its Stargate site in Texas — OpenAI on Thursday released GPT-6 Astra, which president Greg Brockman called a “generational leap” and said could eventually be seen as the arrival of artificial general intelligence, or AGI.
ARC PrizeGreg Kamradt
OpenAI's GPT-6 Astra on ARC-AGI-3 Summary GPT-6 Astra scores 62.7% for $26K on ARC-AGI-3 Semi-Private with our , and 99.9% for $19K with a . GPT-6 Astra surpasses the human baseline in action efficiency on ARC-AGI-3. It used fewer actions than the median tested human on 96% of levels. A key behavior observed in GPT-6 Astra was its ability to turn unfamiliar environments into compact symbolic world models. It represented game mechanics as logical rules and developed its own domain-specific language shorthand to track state and plan actions. ARC-AGI-3 ARC-AGI-3 is a benchmark for studying agentic intelligence through novel, abstract, turn-based environments. Agents must explore, infer goals, and build internal models of environments to effectively plan actions _without_ explicit instructions. You can play ARC-AGI-3 yourself. Your browser does not support embedded video. These environments only contain core knowledge priors and are difficulty-calibrated through controlled testing with hum
Greg Kamradt / ARC Prize : GPT-6 Astra scores 62.7% on ARC-AGI-3 with the standard harness and 99.9% with a new provider adapter harness; Claude Opus 5 scored 30.2%, and GPT-5.6 Sol 7.8% — Summary — GPT-6 Astra scores 62.7% for $26K on ARC-AGI-3 Semi-Private with our Standard harness, and 99.9% for $19K with a Provider Adapter harness.
量子位量子位
刚刚，GPT-6正式发布！OpenAI：欢迎来到AGI时代 闻乐 2026-09-04 05:50:07 来源： 量子位 全球最强C 编辑部 发自 凹非寺 量子位 | 公众号 QbitAI 它来了它来了！！ 刚刚，万众期待的 GPT-6 Astra 和 GPT-6 Astra Pro ，正式发布！！！ GPT-6 Astra是世界上最智能、最协调的模型，为Computer use、Browser Use、软件工程、网络安全、科学和专业工作树立了新的标杆。 AGI时代，还被OpenAI单方面宣布「开幕」了… GPT-6发布会最后，OpenAI总裁Greg Brockman直接甩下一句： Welcome to the AGI era. （欢迎来到AGI时代） 怪不得Claude、Grok组团掉线，原来是Astra启动后扫描互联网，直接把地球上的LLM全灭了—— 奥创（OpenAI版）真来了（doge）。 当然了，OpenAI这次确实完全没低调，训练规模、能力升级全部拉满。 据介绍，Astra是OpenAI历史上规模最大的训练任务，在得州Stargate园区动用超过 10万块GPU 完成了预训练。 它还是OpenAI第一款由前代模型深度参与训练监督的旗舰产品。 好一个老带新，OpenAI的RSI是真转起来了啊…… GPT-6的价格也正式公布，API定价为： 输入：10美元／百万token； 输出：50美元／百万token 相当于 GPT-5.6 Sol的2.5倍，跟刚刚发布的Fable 5.1持平 。 （GPT-5.6 Sol当前官方价为输入4美元、输出20美元/百万token） 基准测试接近「饱和」 GPT-6 Astra这次最核心的变化，是从「回答问题」继续向 「直接完成工作」 推进。 它不只能生成一段文字或代码，还可以操作电脑和浏览器，进入不同软件执行多步骤任务，最后交付可以直接使用的文档、表格、演示文稿、网站甚至工程项目。 至于成绩单…谁看了都直呼离谱，其中三项成绩尤其扎眼： FrontierMath Tier 4 v2： 97.6% ARC-AGI-3： 99.9% （上一代GPT-5.6 Sol是7.8%） ExploitBench： 100% 一个高难数学，一个陌生环境推理，一个漏洞利用，差点全都被它刷到满分。 其中， ARC-AGI-3 是一项专门考验大模
机器之心机器之心
OpenAI Astra，终于来了。 今天，OpenAI 正式发布新一代前沿模型 GPT-6 Astra 。 而 OpenAI 总裁 Greg Brockman 更是毫不掩饰对 Astra 的评价。在发布前的媒体简报会上，他直言自己认为 「我们已经到了 AGI」 ，并以一句：「Welcome to the AGI era.」结束了整场发布。 这款模型早在一个多月前就已进入公众视野。8 月初，《The Information》报道称，Sam Altman 曾在华盛顿向政策制定者演示代号 Astra 的新模型。它支持多个 Agent 长时间协作完成复杂任务，内部甚至一度考虑将其命名为 GPT-6。 但此后，围绕 Astra 的焦点迅速从「有多强」转向了「是否足够安全」。 OpenAI 随后确认，Astra 已达到准备框架的「关键级」网络安全能力门槛，成为公司首个跨过这一等级的模型，并一度放慢部分开发工作、加强安全措施。 昨天，Astra 的另一项变化又浮出水面。据《The Information》报道，它可能引入了 「循环深度」（recurrent depth） 技术，让模型在输出下一个 Token 前完成更多内部计算。换句话说，它可能不仅更强，而且开始真正做到：少说，多想。 经历一个多月的曝光、安全刹车与技术争议后，Astra 终于正式上线。 那么，这次 OpenAI 到底带来了什么？我们先来看看奥特曼最喜欢的宣传视频。 「世界上最智能、最对齐的模型」 首先在官网视觉上，Astra 就已经和 OpenAI 过去发布的模型明显不同。相比此前更简洁克制的产品页，这次 OpenAI 明显强化了「新世代」的发布感：动态变化的数字 6、深色界面，再加上更具冲击力的整体视觉设计，都在强调 GPT-6 Astra 与上一代模型之间的不同。 OpenAI 对 GPT-6 Astra 的定位相当直接： 「世界上最智能、最对齐的模型。」 按照官方介绍，Astra 集中了 OpenAI 多年来在预训练、强化学习和对齐上的研究成果，在电脑操作、浏览器使用、软件工程、网络安全、科学研究和专业工作等领域全面刷新此前模型的能力。 一些 Benchmark 已经高得接近「刷满」。在 ARC-AGI-3 上，GPT-6 Astra 拿到了 99.9%，相比之下 GPT-5.6 Sol 只有 7.8%，
Matt Shumer / Something Big Is Happening : Review: GPT-6 Astra can adeptly use tools like Unreal Engine to build complex environments, such as a civilization with Unreal's autonomous MetaHuman characters — Everyday work, ambitious experiments, and the Manager Loop. … For nearly a year, OpenAI models were my unquestioned default …
Celia Ford / Transformer : OpenAI says it can't read all of Astra's reasoning and admits covert sandbagging would likely go uncaught, yet still calls it the world's most aligned model — OpenAI is hailing its new model as “the world's most intelligent and aligned”, but the details reveal an awareness of being evaluated …
Zac Hall / 9to5Mac : OpenAI rolls out GPT-6 Astra to Pro customers on the $100/month or $200/month plans — Update: A day later, GPT-6 Astra is rolling out to Pro customers on the $100/month or $200/month plan. This follows other new model releases before Plus customers on the $20/month plan gain access.
Hacker NewsTopfi
Simon Willison
I got access to GPT-6 Astra this afternoon, so naturally I used it to generate SVGs of pelicans riding bicycles - at low, medium, high, xhigh and max reasoning levels (Astra doesn't support reasoning=none). Then I rendered those pelicans in a comparison grid with GPT-5.6 Sol, Terra, and Luna, and beyond being fun the result was surprisingly useful. See the grid for full quality images. Here's the transcript that created the GPT-6 Nova pelicans. There are a few interesting things that stand out from this grid. The Astra pelicans are much better . The very best GPT-5.6-Sol pelican (I liked xhigh better than max) is still pretty clearly a bunch of abstract shapes. Every single one of the Astra pelicans, from low to xhigh, looks better than that. The Astra max one is really good. Astra below max still doesn't reliably get the pelican legs on both sides of the frame. In terms of cost, Astra may be around twice the price of Sol ($10/million input, $50/million output, compared to $5/$30 for S
Introducing GPT-6 Astra for developers Blink and you'll miss it, but there's a familiar creature at 1m59s : Across the board, Astra has more attention to detail, better understanding of the user's prompt, and can build more sophisticated outputs. In particular, it excels at building 3D models. I've seen it make incredible renderings of gardens, shipyards, animals , cityscapes, even Dyson spheres. Astra really does believe in putting a red neckerchief on a pelican riding a bicycle. Via Hacker News comment Tags: ai , openai , generative-ai , llms , pelican-riding-a-bicycle , gpt-6-astra
Hacker NewsAnon84
September 4, 2026 A follow‑up to our comparison of Claude Fable 5 and Fable 5.1. We gave OpenAI's GPT‑6 Astra control of the same YAM arms under the same Inspect Robots agent policy, on the same two tasks: “Pick up the red block from the table and place it inside the bowl.” “Pick up the round blue puzzle piece by the knob at its center and place it into the matching circular groove in the board.” On the bowl task Astra placed the block in **19 of 20** trials, against Fable 5.1's **8 of 20** and Fable 5 in 1 of 20, in 2.5 minutes per trial to Fable 5.1's 6.8, at an estimated $0.94 per run to $2.12. The puzzle task is a different story: Astra completed the insertion 2 times in 20 against Fable 5.1's 2 in 20. It reaches the groove and stalls at the same final step Fable does, at $1.36 per run to $2.18. Video 3 Block into bowl: the best completed run of each model (highest stage, then shortest), each played in its own time at the same speed‑up. Timers show real elapsed time with thinking p
Emily Forlini / Fortune : OpenAI quietly updates its evaluation metrics for GPT-6 Astra, making changes that appear to favor Astra and continuing to revise other metrics after launch — OpenAI has changed several evaluation benchmarks for its GPT-6 Astra model since first publishing a blog post announcement mid-afternoon on Sept. 3.
编辑｜张倩 最近两天，相信大家都被 GPT-6 Astra 的各种逆天玩法刷屏了 —— 一句话完成 3D 建模、从零制作完整游戏、几千个零件的东西说拆就拆，甚至还能拿来控制机器人…… 总之，强得没边。 网友们有句话说得很夸张：只要你学得慢，你就可以不用学。但是 GPT-6 Astra 确实给人一种感觉，它一出来，原来困扰很多领域的复杂性突然就有解决的苗头了。 这种趋势不容小觑，但模型现阶段的能力也不宜过度夸大。在这篇文章中，我们就来盘点一下 GPT-6 Astra 的各种逆天玩法，及其当前的局限。 3D 建模这件事，突然开始变简单了？ 这两天 Astra 最密集刷屏的能力，大概就是 3D。 有人拿它来直接生成 3D 的火车： 两列火车都是 Astra 用 TypeScript 和 Three.js 代码，在浏览器运行时直接生成出来的。车身尺寸、轮廓、各种几何结构来自代码里的尺寸参数、profile 和 geometry function；车轮运动是代码，整辆火车炸开、零件散开再重新组装的动画也全是代码。 而且，这些模型做出来之后还能按部件拆开。一台相机可以被拆成 122 个组件组、1877 个独立建模部件。整个任务大概跑了 3 个小时，作者表示之前根本没意识到 Three.js（一个用于在浏览器中创建和展示 3D 图形的 JavaScript 库）还能这么玩。 相机能拆，特斯拉也能拆。有人将其拆成了 334 个建模部件，而且这 334 个零件的名称、编号和基础结构是有官方依据的，不是 AI 凭空捏造的。当然，它仅仅是一个非常粗略的框架级拆解，远远达不到 3D 拆解所需的精细度。 有意思的是，这个作者拆完特斯拉不过瘾，还建了个网站来「拆人（男性）」：全身的 2234 个建模部件可以逐一查看，而且可以映射到源文件。 看到这些案例，有人高呼「3D 建模的学生天塌了」。 但也有人说，AI 现在直出的结果还不能商用。 而且，稳定性也是一大问题，它终究还是需要人来修 bug 的。 这些 Demo 的意义，并不只是证明 Astra 会建模。更重要的是，它展示了一种新的交互方式：人不再需要学习复杂的 3D 软件，而是直接描述想要的世界，让 AI 负责把想法翻译成空间结构。 过去，3D 建模是进入数字世界的门槛；未来，它可能变成 AI 理解和生成现实世界的一种基础能力。当然，距离真正
