Keep on to blur preview images; turn off to show them clearly
also indicative of later phase where models explore more diverse semantic paths.
Reasoning models coming (very) soon. Co-founder @pleiasfr
TrustMRR > DataFast Funnel: 👋 1,012 unique visitors ❤️ 36 new users 💰 +$279 MRR Most users are still under the 14-day free trial. Not bad for a 24-hour side project with an iframe.
🧑💻 https://t.co/Y30jsaHwz9 $20K/m ⚡️ https://t.co/vatLDmi9UG $17K/m 📈 https://t.co/3EDxln5mdi $16K/m ⭐️ https://t.co/MZc8tG9xWi $8K/m 🧬 https://t.co/SfrVXVtmdA $.5K/m 🍜 https://t.co/r07EpGSYJ2 $0K/m 🧾 https://t.co/7olaOzV8Xd $0/m +18 https://t.co/4zCWHGJp1S
今天拿 k2-thinking (cc) 和 gpt-5 high (codex) 尝试修了一些业务问题。 说 2 个细节上的差别 1. 速度上，k2-thinking 明显快一档。我习惯拆成「定位」和「修复」两步，它在第一步特别利索（不过其他家也比Codex 快很多。。。。。） 2. 提示词敏感度，Codex 的 review 更「自觉」，不用多提示就能顺藤摸瓜，挖出代码一致性这类深层问题，顺便给一堆建议。k2-thinking 则需要多盘几句才能激活——比如追调用链 bug 时，加功能可能没啥问题，但前置的逻辑不一定能同步，但逼问「这整个链条里面还有什么需要注意的吗？」就可以有比较好的效果。 比较惊喜的点是，一旦让它盘整条链路，它会主动标记疑似问题点再逐一排查，这个体感非常不错。（图 2）
🖥️ Indie Maker 🛠️ 星球「海哥和他的小伙伴们」 📌 油管「海拉鲁编程客」 🌸 沦为程序员的段子手/猫咪
The only sensible thing @tegmark has said since he started talking about AI risks.
AI @amazon. All views personal!
生活的经历还告诉我另一个道理：自证是可怕的消耗，不必理会无谓的诽谤。
🔖 Building · Sharing ✨ Zipic · TimeGo · KeygenGo · HiPixel · Orchard · Prism
The algorithm is like a lottery. More posts. More changes of getting a winning ticket.
I build stuff. On my way to making $1M 💰 My projects 👇