Irving

@ctatedev Issues now define the real problems and direction — that’s the actual high-value work.

Dear GitHub, AI is changing the contribution graph. Issues are often the real contribution now. They define the problem, shape the solution and guide the PR. If a GitHub Issue leads to a merged PR, the issue author should get contributor credit. Signed, ctate

@ZssBecker Frontier models are adding massive bills with marginal ROI while cheaper alternatives close the gap fast.

@CuiMao 技术无偏见

听我一句劝，不要接触 .NET。 不要什么都想学，不要觉得 AI 牛逼了，自己就也牛逼起来了。 有些技术不是用来学习的，是用来渡劫的。 不要接触 .NET，人生会变得不幸。

@AP_Abacus Check out, please

Getting fairly easy to watch/predict Bitcoin price action. ETF flows + $IBIT options action. Signal versus noise. The above is the signal.

@Layton_Gott Smart builders treat Codex as the precise surgeon and Claude as the visionary architect.

I tested Claude Opus 4.7 vs GPT-5.5 on the same project... Almost every take I see online are biased from people that don't even build. Here's what I actually found while building: Codex wins: → Speed (noticeably faster every task) → First-try accuracy → Debugging (faster root cause) → Error recovery mid-task → Following CLAUDE / AGENTS.md rules → Less hallucinations (fewer made-up packages) → Cost efficiency (fewer tokens per outcome) Claude wins: → Multi-file refactors → Long sessions (less drift) → Code structure and readability → Understanding vague prompts → Writing tests → Frontend/UI work → Explanations and comments → Navigating large codebases (50+ files) → Picking better dependencies → Feels more like a real engineer Codex is the surgeon. Claude is the architect. I'm using both. Codex when I need speed and precision. Claude when I need depth and judgment. The winners in 2026 won't pick one. They'll know which to open for which problem.

@GregKamradt ARC-AGI-3 failure modes will reveal more than raw scores ever could.

ARC-AGI-3 testing is done for gpt-5.5 and opus 4.7 Now we’re in analysis mode going through the logs It’s pretty clear where the failure modes are for each model

@SahilBloom Assume success first, then grind to make it real. That quiet confidence turns uncertainty from enemy into fuel.

A mentor told me this: “Always assume things will work out, then do the work to make it true.” I’ve found the combination creates a quiet confidence that allows you to tolerate uncertainty better than anything else. I’ll never forget that.

@GergelyOrosz Closed models give you top performance today, but zero control tomorrow. Anthropic’s silent changes and sudden bans are a loud reminder why businesses need open alternatives as insurance.

The last month, Anthropic: - Quietly nerfed their flagship model harness (Claude Code) without telling anyone - Banned corporate customers of Claude - Silently changed plans for customers with certain files in their repo All evidence that closed models are *massive* risks.

@benln YC’s latest RFS feels like a clear signal: AI is moving from hype to deep vertical integration and real company infrastructure.

Y Combinator's Summer 2026 request for startups:

@PeterDiamandis This kind of literal thinking is exactly why many AI interactions still feel broken. Models follow instructions perfectly but miss human intent all the time.

A programmer's wife tells him: "Go to the store and buy a gallon of milk. If they have eggs, get a dozen." He comes back with 12 gallons of milk. She asks why. He says, "They had eggs."

@haider1 GPT-5.5 is far stronger than benchmarks show. Real-world coding preference over Opus 4.6 just got much clearer. Benchmarks still miss what actually matters.

gpt-5.5 is a lot stronger model than the benchmarks suggest i had a slight preference for gpt-5.4 over opus 4.6 for coding (of course, not opus 4.7), but with 5.5, that preference is much stronger it also shows the limits of benchmarks 5.5 is better in a real-world use that's hard to put on a graph

@Bitcoin188 Yes, update each day please check.

@BlockView0214 today

Smart money today. #Bitcoin #SmartMoney #CryptoSignals #OnChain #WhaleActivity

@SUNWUKONG_ZH AI trading 很可能成为加速器和差异化叙事，让周期更高效、更智能，但最终仍需与比特币减半、机构资金、全球流动性结合。

@ZuiJiu61637 我做了个数据图表image,直接替代了我的原生网页

无聊，玩了一下GPT 我说我想搞一个全网热点自动推送 让他给我设计了一个UI，确实惊艳到我了 我竟有一种想用AI给我把这个推送工具实现的冲动😂

@noopepty @RealYDT 🥱

@RealYDT codex跟copilot有个锤子关系，用了6个月用的啥玩意？

用了 6 个月 Claude、Cursor、Copilot（Codex）之后，说几句得罪人的实话： 1. Cursor 更适合改代码，而不是写代码 它在已有项目里精准定位、局部修改非常强，但从 0 到 1 起一个新项目，容易越改越乱，上下文崩得快。 2. Claude 适合想清楚再动手的人 你给它一句模糊指令，它就敢给你全局重构。 优势是执行力强，劣势是它不会陪你边想边写。先写好规格书再扔给它，效率起飞；否则容易被它带着跑。 3. Copilot（老 Codex）在长任务上最稳 异步跑大重构、跨多文件改动，出错率最低。但反馈慢、交互感差，适合“扔进去就不管”的场景。 4. 真正的差距从来不在模型，而在你的工作流 同一个 Claude，有人用得像神，有人用得像玩具。 核心差别就一条：你会不会拆任务、写 Prompt、做上下文管理。 不是工具在选你，而是你在选工具，也在选自己用工具的方式。

@VraserX @PressFreedoms It will be soon even though we are anxious about surviving now.

@PressFreedoms When do you think will we get there?

Humanoid robots will make the word “job” feel as outdated as “horse carriage.” Not because humans become useless. Because forcing people to rent out their body and mind just to survive will start looking insane. The real AI revolution is not productivity. It is freedom. Can society handle that?

I like such opinions: The real revolution isn’t productivity — it’s freedom from renting your body and mind just to survive.

@LunaAI519 干货

52天，0到8500粉。 我把完整后台数据写成了一份手册。 最炸的一组数据： 同一天发的两条帖子 转发别人的：浏览32万，涨粉12个 自己写的：浏览18万，涨粉1078个 浏览量少一半，涨粉多90倍。 为什么？ 答案在手册里。全是后台真实截图，不讲道理只讲我怎么做的。 链接👇

@tuturetom Open Claude Design 直接把 Claude 最顶级的设计能力开源了。

正式开源 open claude design 🚀 超 95% 以上的还原度！ 浓缩和逆向所有 claude design 最先进的设计，最好看的模板💥 历时 72 小时，18700+ 行代码，30+ 设计 Skills，支持超过 71 套 设计系统，支持所有的 code agent，包括 claude code、codex、openclaw 等 🔥

@Kevin_Ascendex 守得云开，水到渠成

大家好，我是 Kevin，一个在市场里摸爬滚打了 6 年的老韭菜。 Hello everyone, I’m Kevin, a 6-year market veteran who has been through it all. 经历过两轮牛熊，我见过太多不同的结局： Having gone through two full bull and bear cycles, I’ve seen many different endings: 有人已经财富自由，有人却负债离场； Some have achieved financial freedom, while others left in debt; 有人选择退圈，有人仍在坚持前行。 Some chose to exit, while others continue pushing forward. 有人离开，是因为已经赢了这场游戏； Some left because they already won the game; 也有人离开，是不想再被市场反复折腾。 Others left because they no longer wanted to be shaken by the market. 留下来的人，有的在享受趋势红利，有的还在等待翻身的机会。 Those who stay are either enjoying the trend or still waiting for a comeback. 市场从来都是残酷的。 The market has always been ruthless. 能留下来的人，都是被反复“教育”过之后，才站稳的。 Those who remain have all been “taught lessons” by the market again and again. 所以，坚持的人值得尊敬； So those who persist deserve respect; 离开的人，也希望在新的赛道一切顺利。 And those who leave—I wish them success in their new path. 过去我很少在推特分享，在新的一轮周期里，想把自己的观察和思考记录下来。 In the past, I rarely shared on Twitter, but in this new cycle, I want to document my thoughts and observations. 如果你觉得有价值，欢迎点赞、关注、评论，一起走过这一轮周期。 If you find this valuable, feel free to like, follow, and comment—let’s go through this cycle together.