每天重置一次,让 Anthropic 远离。他们心里清楚。https://t.co/1u7mpAd2Xt
What the frick: one benched codex reset for every day we don’t have access on our paid plans? Absolute legend. Our boy never disappoints. Kudos OpenAI. Best community work. https://t.co/USBkOakeUL
@kimmonismus · X
每天重置一次,让 Anthropic 远离。他们心里清楚。https://t.co/1u7mpAd2Xt
What the frick: one benched codex reset for every day we don’t have access on our paid plans? Absolute legend. Our boy never disappoints. Kudos OpenAI. Best community work. https://t.co/USBkOakeUL
GPT-6 Astra 在 Epoch AI 的能力指数(ECI)上取得 169 分,刷新此前 163 分的纪录。


GPT-6 Astra has set a new ECI record, with a score of 169. This is a substantial jump from the prior best (163), but is within our uncertainty range for the reasoning-era ECI trend. Astra also set new records on our math, continual learning, and game-puzzles benchmarks. On our long-horizon coding benchmark, MirrorCode, Astra ranks between Opus 4.7 and Fable 5. OpenAI gave us pre-release access to test Astra. Charts and more details for Astra’s individual benchmark results in the thread.
推荐理由:借 Epoch 与 Artificial Analysis 两套指数的分歧,可以看清单个能力纪录与综合实用性评价之间的差距。
https://t.co/DzyFtV41Nh • https://t.co/82Q0SvApLB 如果你自己也跑了这个相同的测试,我真的很想看看你的结果对比如何。
作者用相同提示词、参考图、时长且不做剪辑的方式对比了 HappyHorse 1.1 与 Kling 3.0,从唇形同步、角色与场景一致性、复杂运动、镜头控制和价格工作流五个维度逐一测试。
We will give one banked reset for every day you don't have access to Astra on your paid ChatGPT plan, starting today. Team is moving mountains to give access as fast as we can. First one will land in ~ 3 hours. There is still time to create your account if you don't have one.
今天的教训:我们急需新的基准测试。 (引用推文:arc-agi-3 现已饱和)
arc-agi-3 is now saturated https://t.co/sD4ePv5cws
我想我明白为什么他们叫它 AGI 了。它确实是通用智能。它的能力在各方面都真正出色。它就是什么都擅长。https://t.co/aUqDaEZsff https://t.co/fKKK1Mu8Aa




OpenAIs website is finally live and running. And you know whats insane? GPT-6 Astra is not even OpenAI last model that will come out this year. Fable 5.1 crushed today. Good luck with that IPO. Maybe its time to increase our weekly rates https://t.co/kyZxekMokV https://t.co/fOSiiln6Gb
You can quote me on this: OpenAI crushed Anthropic’s IPO. They put the newly released, state-of-the-art Fable 5.1 to shame. And they knew exactly what they were doing. Just look at the numbers. Its not even close. https://t.co/pMgJwWhm4g https://t.co/ktkPEvwMHX
GPT-6 完整官方发布片段。不知道 OpenAI 现在为什么把发布会搞得这么糟糕。 https://t.co/HdZtatYQ0e
Official GPT-6 Astra Benchmarks from OpenAIs website "Astra also saturates ARC-AGI-3 with a 99.9% score and ExploitBench with a 100% score" "GPT‑6 Astra is rolling out today to a limited set of organizations and over the coming days will become available to all ChatGPT Plus, Pro, Business, and Enterprise users, as well as through the OpenAI API and AWS." Insane jump. AGI is here.
推荐理由:作者比较了 GPT-6 Astra 与 Claude Fable 5.1 的基准成绩和 API 成本,可据此看性价比差异。
你可以引用我的话:OpenAI 碾压了 Anthropic 的 IPO。 他们让刚发布的、最先进的 Fable 5.1 相形见绌。而且他们完全知道自己在做什么。 看看数据就知道了。根本不在一个量级。
Full benchmarks. Absolutely insane. ARGI-AGI 3, from 7.8% to 98.6% https://t.co/0F1Vdrhpve https://t.co/KsgD7QmD7I
完整基准测试。简直疯狂。 ARC-AGI 3,从 7.8% 提升到 98.6% https://t.co/0F1Vdrhpve https://t.co/KsgD7QmD7I
GPT-6 Astra Benchmarks: FrontierMath Tier 4 v2: 97.6% DeepSWE v1.1: 74.1% BenchCAD: 95.9% GPQA Diamond: 96% ExploitBench: 100% ARC-AGI-3: 98.6% (reported) Access in "plans to bring the new model to paying ChatGPT subscribers and its API over the next several days." https://t.co/sOc2OaI3qL https://t.co/HHmIHksETD
GPT-6 Astra released. Greg Brockman: This might be the arrival of AGI. "Generational leap" https://t.co/2E9A3PYDdt
GPT-6 Astra released. Greg Brockman: This might be the arrival of AGI. "Generational leap" https://t.co/2E9A3PYDdt
GPT-6 Astra 发布。Greg Brockman:这可能是 AGI 的到来。“代际飞跃” https://t.co/2E9A3PYDdt
Codex 里已经有两个新模型了: GPT-6-Astra 和 GPT-6-Astra-Aeon 随时可能上线。
到这一步,我可以说已经确认 GPT-6 今晚发布。就等直播公告了。 我准备好了!https://t.co/HUqhoqOTqF
一份新预印本证明,四名参与者、最多 9 件物品的公平分配存在 EFX 完全分配,此前四人情形的通用保证止于 7 件物品。
Claude 宕机了。OpenAI 可以做件最搞笑的事,趁这个时机来个惊喜发布(*Astra*)https://t.co/pp8QykrUAN
推荐理由:这次开源同时公开训练代码、数据配方、checkpoint、日志与评测,读者可据此查看模型从训练到结果的完整记录。
NVIDIA 将收购 Hugging Face,Jensen Huang 表示 NVIDIA 会成为 Hugging Face、其社区以及开放模型未来的好归宿。


Exciting day for NVIDIA and @huggingface. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. They allow every developer, startup, university, industry and country to build with, customize and benefit from AI. Thank you @ClementDelangue for coming to me. NVIDIA is going to be a great home for Hugging Face, its community and the future of open models. 🤗 https://t.co/q8Om2Xc5ye
推荐理由:作者从 NVIDIA 的商业利益出发解释这笔收购为何可能利好开源生态,给出一个不同于担忧垄断的判断视角。
Meta Muse Spark 1.3 现已在 OpenCode 上免费开放。转发这条消息的 @kimmonismus 称智能便宜到无法计量,并表示自己喜欢这种竞争。
Meta Muse Spark 1.3 is now free on OpenCode
更多证据表明 GPT-Astra 已经非常接近了 OpenAI 的一篇帮助文章几小时前刚更新。看起来他们已经在为发布做一切准备了。
OpenAI just exposed a public branch containing new JavaScript and Python computer-use environments explicitly labeled “gpt6.”! This reinforces the indications that GPT-Astra is, in fact, GPT-6. Astra is so close, I can almost smell it!