Dumbo

@DevMunguia @paulveth @Yampeleg Me nether. FFFFFu.

@paulveth @Yampeleg Yeah I was working when it gave me error 500

opus 4.7 today or tomorrow?

@moran_neko @hsn8086xe 难绷🤣

@ytx_1999 @hsn8086xe 英语老师念咒念出来的

Hurry up and release opus 4.7 already.

Real talk — Apple Silicon optimization is moving fast, but the compute gap with NVIDIA? Still massive. You can optimize all you want, the hardware ceiling is the hardware ceiling. Maxed-out Apple still doesn't touch NVIDIA for local LLM stuff. And dense models like Gemma 4 31B and Qwen 3.5 27B absolutely destroy their MoE variants in inference. Raw GPU power matters more than ever. So yeah, NVIDIA GPUs = still the move for serious local LLM work. Don't bet on Apple closing that gap anytime soon. …anyway I ordered a 128GB M5 Max on April 5th 🤡

@ytx_1999 You are right! I believe that Apple Mac mlx still has a lot of potential for future improvement, and it may even squeeze more performance out of older devices.

M5 Max MacBook Pro or RTX5090 for Qwen3.5 27B?😵‍💫🤫🫠

@iamkuni @ethanhuang13 哥们M1现在还没换，前几天刚下定的M5，1-2周送达，急不可耐了已经🤣

@ethanhuang13 生活能夠簡單就是一個看起來很簡單做起來很難的一件事。 這麼說起來，我M1也是一直撐到去年底才換。

iOS 工程師職涯十來年 經歷過 3、4 台頂規 Mac 用來開發是很爽 但我現在覺得 如果生活能夠簡單到只需要一台 MacBook Neo 就好了 #週間廢推

@trq212 😎

/buddy

@claudeai Cooooooooool!!!

Computer use is now in Claude Code. Claude can open your apps, click through your UI, and test what it built, right from the CLI. Now in research preview on Pro and Max plans.

@iruletheworldmo @grok is opus 4.7 coming?

a little birdy keeps chirping in my ear opus 4.7 opus 4.7 go away little birdy.

@milkblue1 @geekplux 补一下，已经换cmux了，实测更好用🤣

@milkblue1 @geekplux 1.3.0是有这个bug，很烦，我就卸载了。然后我看官方1.3.1修复了我又装回来了，现在用着没问题了。

之前一直用 iTerm 配 oh-my-zsh，今天刚下载了 Ghostty，是不是跟上时代了 🤣

@NyshaDev I'm 26. Let me introduce myself. - 2 years of frontend work experience. - 2 years of Flutter & backend work experience. - Claude Max 20x subscription user. - Trying to build some products by myself with AI. IF YOU'RE INTERESTED IN ME.. LET'S CONNECT.

I'm 22 a software developer. I want to connect with people who love:- - Coding - vibe coding - Full stack developer - software engineer - AI/ML IF YOU'RE INTO TECH.. LET'S CONNECT

Claude Code v2.1.75 just dropped 1M context for Opus 4.6 — now included in Max subscription, no extra usage needed. Gone are the days of constant compacting at 200K. A full session without losing context. Finally. 5x more room, same pricing.

@claudeai Always be your fan!!! 🥳🥳🥳

1 million context window: Now generally available for Claude Opus 4.6 and Claude Sonnet 4.6.

@0xInk_ Nice choice.

just switched from ChatGPT to Claude

@gailcweiner Absolutely right.

GPT 5.4 is absolute shit.

If you're interested in the exact setup: - Vault structure (PARA method) - CLAUDE.md template - Command definitions - Tool chain config Drop a comment and I'll share the details.

The shift that changed everything: Stop treating AI as a tool you prompt. Start treating it as a teammate that owns systems. Give it persistent memory (vault). Give it clear boundaries (CLAUDE.md). Give it autonomy (commands + multi-tool access). Then get out of the way.

Most people use Claude as a faster ChatGPT. I use it as my second brain that autonomously manages 11 parallel projects. Here's the architecture that lets one person do what used to take a team: