Hekate retweeted
OpenAI’s @sama joined me at the pub (and had his first Guinness) to discuss the PHP exploit, math weekend at OpenAI, agent swarms, and how he thinks AI leads to deflation.
Hekate retweeted
holy shit,不愧是谷歌大善人😭
Google 这是要直接把教育出版和考研辅导给干翻啊,
几百本大学正版教科书直接免费内置,一键挂进 Notebook 当知识库随时提问,学生党这下连买书的借口都没了😂
Hekate retweeted
🇺🇸 Coffee in one hand, phone in the other, feet up, and the Tesla Cybercab does the driving.
Elon is really trying to turn rush-hour traffic into ME TIME.
Honestly, I could get used to this 😂
Writer: Sol
Hekate retweeted
The Grok Bot wave. Massive growth in just a few days.
Why?
It is the easiest most accessible agent platform thus built, and it keeps getting better,
I am not guessing, I have been using agent systems before we had LLMs and today I can use any bot, but I find @Grok @Bot to be the best on all my test points.
It has not retreated from anything I threw at it. But that’s only half of it. The suggestions are great and unlike any other platform.
The chart is a wave and I say catch it…
Hekate retweeted
JUST IN: Anthropic recorded a staggering $80,000,000,000.00 net loss in the first half of 2026.
Hekate retweeted
❗️ A Mac user says Claude Code was building them a temperature monitor app when it found a cryptominer using 12 cores.
The readings looked too high, so it checked the CPU and found XMRig at ~1175%, with the chip at 112°C.
Hekate retweeted
We keep improving performance across Claude, and you may have noticed it starts responding sooner!
Claude Code now gzips the conversation it sends on every turn, so it reaches the model faster. Cloud sessions also start up as soon as you start typing.
Grok 推送的 AI 资讯还挺好的,但你可能想要在其他 Bot 或者 Agent 使用,一个简单的方法就是去问它是怎么做到的:
> 你在帮我整理新闻时,是怎么去X检索我想要的资讯的?怎么判断资讯是不是新是不是有价值的?
其实原理也很简单,就是关注一些特定账号,分三类:AI 实验室和公司官方号、关键人物、新闻和爆料号。
也列出了能直接复制的搜索语句,覆盖模型发布、产品和 agent、融资和收购、监管和安全、爆料、中文资讯,还有我自己关注的方向。
另外写了怎么判断新旧、怎么判断有没有价值,以及推送格式。
再让它把这个流程整理成一个 Skill,就能去复用这个流程了。
不过最关键的是,其他 Bot/Agent 你得自己花钱用 API,而 GrokBot 的 X API 不用额外花钱。
Grok 整理的完整提示词我放评论。
完了完了😱中国模型现在压力更大了!!!
Haiku5.5 已经拿Flash的价格,
杀进中国模型旗舰区了 😱
43 分,输入 $0.10,输出 $0.50
只高 1 分的 Kimi K3,输出贵 30 倍
高 2 分的 GLM-5.3 Max,贵 8.8 倍
高 2 分的 Qwen3.8 Max,贵 12 倍
高 3 分的 MiMo-V2.6-Pro,也贵 1.74 倍
再往下看就更离谱!!!
低 4 分的 DeepSeek v4.1,反而贵 2.4 倍
低 9 分的 Qwen 27B,贵 6 倍
低 14 分的 MiniMax-M3,也贵 2.4 倍
Haiku 5.5 已经不是便宜的小模型了。
它现在最恐怖的地方是:
高并发、高频调用、批量跑任务,
拿 Flash 的成本,干接近中国旗舰的活!!!































