JinHan

💼 15 年产品人 | 前阿里高级产品专家
🤖 Ex 网易云 / 猎豹移动 / 某A轮 AI 公司产品合伙人
📊 亿级电商&内容平台 | 0→1千万DAU工具App
🔥 落地AI产品中｜分享实战经验｜重新出发
🔗 个人站点： 优先回复订阅者评论&私信

[JinHan @AIJinHan]

Stop joking. The production quality of these videos is roughly at the level of Chinese model companies back in January, and not even the top-tier ones at that. Following the release of Seedance 2.5 and MiniMax H3 at the end of July, the overall production quality has taken a massive leap forward. Just recently, China's first AI anime series is officially landing on a well-known TV channel in their country for broadcast. Anyone interested can go have a look.

[JinHan @AIJinHan]

@Pluvio9yte 今天还跟朋友聊到过这个话题。如果想要流量就别讲太深的东西，因为绝大多数受众可能都是初级，讲太深的东西反而没流量

[JinHan @AIJinHan]

好消息是硕果仅存 Google 的 Gemini 3.8 Flash，这是老美的算力中心，光缆被挖断了，还是拉闸断电了？坏消息是，你的中转站居然都还能跑

[JinHan @AIJinHan]

@huangyun_122 然后帐号过不了多久就收到站内封控提醒，再过会就封号了

[JinHan @AIJinHan]

@xingbugengming 和 OpenCodex 那个项目比，哪个更好用一些？

[JinHan @AIJinHan]

在你能自己掌握的事情上，以及自己想清楚的事情上，他们是很好的执行者

[Yihui @yihui_indie]

AI 真的能帮我省下时间的事情，都是我最有掌控感的事情。

[JinHan @AIJinHan]

@loki_yan_seo 表头和框架描述。哈哈哈

[JinHan @AIJinHan]

普通人更需要共情的是乱世尘埃里面的皑皑白骨，而不是帝王将相。近期位于山西运城市稷山县的玉璧城遗址，因为强降雨后露出更多东魏士卒遗骸白骨，抖音里一堆人在这里共情高欢的遗憾，真置身那个时代，有多少人会是韦孝宽、斛律金，而又会有多少人会是被裹挟冲阵的普通人

[JinHan @AIJinHan]

要人味点，可以用豆包或者gemini

[Loki Yan SEO @loki_yan_seo]

朋友们，目前中文写的最好的AI是什么？

我想找一个AI帮我加工一下我的文笔🥹🥹

我这写作水平，当年在班里属于差生

[JinHan @AIJinHan]

如果本身有 Data for SEO 的 API 的话，理论上这层皮没必要用

[Ihtesham Ali @ihteshamali]

Semrush charges $139.95/month.
Ahrefs charges $129/month.

A guy named Ben built the open-source alternative that you can run on your computer.

It is called OpenSEO.

Enter a competitor's website and it reveals the keywords bringing them traffic, the pages winning those searches, and the backlinks helping them rank.

And enter your own site and it tracks rankings, crawls technical problems, reads Search Console, and shows whether ChatGPT or Google AI Overviews mention your brand.

Then connect Claude or Codex.

Your agent can pull the real numbers, compare opportunities, cluster keywords, and save a strategy back into the dashboard instead of guessing from a prompt.

The hosted version starts at $10 a month. The entire codebase is MIT licensed, so you can also self-host it and pay DataForSEO directly for only the data you use.

No, it does not magically reproduce every database and advanced feature Ahrefs built over a decade.

But it breaks the most important rule of the old SEO market:

You no longer need their permission to own the tool.

https://t.co/WU9wr98RNo

[JinHan @AIJinHan]

@jacekdziwisz Exactly. With local LLMs, the model has to fit before speed even matters. 512GB of unified memory turns the M5 Ultra into a very quiet server under your desk — minus the mini power plant 😅

Full disclosure: I built a small Mac/model fit checker:
https://t.co/Eb2X7urpQE

[JinHan @AIJinHan]

こういう「Mac mini Air」枠、確かに欲しいですね😄 ただ、安いMacほど「何ができるか」より「何をしないか」を先に決めないと、後からメモリ不足が一番高い買い物になりそうです。用途から「買う・待つ・維持」を考える無料ツールも作っています（英語・自作、Appleとは無関係）：https://t.co/lDklAiu6C9

[JinHan @AIJinHan]

これは「載る」と「使える」が別物だとよく分かる実測ですね。prefill 446 / decode 23.2だけを見ると快適そうでも、32kで10tps未満ならAgentや長文処理では別評価。M5 Ultraは帯域で改善余地がありますが、同じruntime・build・contextで再測定しないと「解消」とは言えなさそうです。モデル適合・context・runtimeを分けて見る無料ツールも作っています（英語・自作、Appleとは無関係）：https://t.co/F944v7H8N1

[JinHan @AIJinHan]

@lnkiai これまでにどのようなプロジェクトに取り組んでこられたか、教えていただけますか？ 得意な分野や関心のあることを把握できればと思います。

[JinHan @AIJinHan]

32Bが「普通に使える」速度になると、論点が「ローカルで動くか」から「どの仕事を任せるか」に変わりますね。RTX 5090はVRAMに収まるモデルの速度が強い一方、Macは大容量ユニファイドメモリと静音性が強み。クラウド不可の開発では、モデル性能だけでなくtool calling・context・継続運用まで見たいところです。Mac向けの条件比較ツールも作っています（英語・自作、Appleとは無関係）：https://t.co/F944v7H8N1

[JinHan @AIJinHan]

Please note that this post may be misleading. Five Macminis wired together might look like they would get a massive boost in computing power, but in reality, whatever you can run on five Macminis is mostly just open-source small models, and their overall performance falls well short of Anthropic's Claude running the latest Fiber 5 model.

[JinHan @AIJinHan]

@MacRumors A discount is great—unless it locks you into too little unified memory 😅 For local LLMs, model size, quantization, context and runtime matter more than the chip name. I built a free planner (independent, not affiliated with Apple): https://t.co/NcCWblPFG6

[JinHan @AIJinHan]

@tom777kruise Expensive yet practical bespoke furniture in China rivals the luxury price tag of a full set of mahogany furniture.

[JinHan @AIJinHan]

对某个小产品多语言一直不太满意，今天进行多语言的优化，意外地发现 Gemini Flash 3.7在这一块上面还挺好用的。做多语言的朋友可以试试！