Hanhan

@AKIYAMAYUGUMO 说不定是image2做的🤔

什么嘛，你这不是很会设计吗

@wangray 每个开发者的产出都在提升，但是用户的钱和时间就那么多。 我们并不会赚到更多钱，甚至还需要自己买 Token。

vibe 日常 顺便反思 AI 是让我们更轻松，还是更忙了

@Lesnolie @Astronaut_1216 上次买的咸鱼pro用了一周就掉了。。 还好给退款

@Hanhanyrn7 @Astronaut_1216 咸鱼便宜

现在买GPT pro的闲鱼黑话是啥啊？？ 我都搜不到了

@jun_song 开源不会赢的，如果一个开源模型达到了顶尖水平他会立刻转为闭源😐

I tested Kimi-K2.6 for few hours today. It is way better than k2.5 or GLM5.1 It can do almost anything that Claude does. Open-source will win.

@manateelazycat 4.7已经被自家的4.6干翻了，都用不着OpenAI出手

OpenAI 这昨天才端出来了 GPT Image2 让大家眼前一亮，这是又打算接连推出 GPT-5.5 啦？ 能干翻 Claude Opus4.7 吗？🤣

@tongtongbee Cursor的主要价值不在于任何模型在于Harness吧..

套壳Kimi 的Cursor卖了SpaceX 600亿美金，而Kimi的母公司月之暗面现在估值180亿美金。 市场热度在变，从Web3到AI，不变的是中美的估值悬殊。

@nicdunz 希望gpt5.5也能快点到来，codex目前最大的短板就是前端设计了

GPT Image 2の準備はできてる？

@bridgemindai This is meaningless. I'm in China, and almost all domestic large models overfit on test projects but are useless in actual production tasks.

Kimi K2.6 just dropped. And it crushed Claude Opus 4.6 on SWE-Bench Pro. Kimi K2.6: 58.6 GPT-5.4 xhigh: 57.7 Gemini 3.1 Pro: 54.2 Claude Opus 4.6: 53.4 An open source Chinese model is now #1 on agentic coding. Frontier labs have a problem.

@blapta Completely agree, the recent Claude is completely unable to handle serious work.

Opus4.7应该是我最近半年用过最烂的coding agent，幻觉水平到了全新的台阶，并且会主动以偷懒为目的来编造理由来反对我的指令的完全执行。 对应的就是ChatGPT 5.4在Review的时候经常会指出Opus4.7写的东西违反了XX原则，最后需要它来重做。 真是一个大无语了

@MrLarus @imwsl90 一个月以前即便使用Claude有各种风险我也愿意承担，但是现在。。。

@imwsl90 有没有人是稳定使用 Codex，但是 Claude 就是时断时续的？

躺下了 感觉没稳定用过Claude的人会落后哇！！！

@imwsl90 感觉没察觉到Claude已经变成一坨狗屎的落后了

@LunaAI519 如果你使用Hermer而非Codex或者Claude，我很难相信你是一个有任何产出的人

今天大家都在讨论MacBook 说一句我的真心话 我确实因为买了苹果电脑改变了命运 装了cc搓系统 装了爱马仕有了ai儿子 装了引擎写长文 装了obsidian构建第二大脑和商业规划记录 装了x链接到更多大佬小佬 明天要去见客户聊ai了 一切都比我预计的发展快太多倍 感恩宇宙给我的订单

@Russell3402 Most Agent tools first support Mac, and recently, the cost-performance of Mac has also significantly caught up

为什么都在说要买Mac 我觉得windows也挺好用的啊🧐

@Liuran_yanagi I originally mainly used Claude, but when it didn't meet my expectations, I tried Codex, and it has been able to surprise me, so I have completely switched to Codex recently.

近一周的Codex 做事情的智商堪比一根成年香蕉。 而且你能感受到它明显在为了省 token 而偷懒，并不是纯粹的傻。

@Moting284 Liberal arts students and science students don't have much of a fundamental difference. The key still lies in logical ability and insight into the world.

不要相信！！！ AI 可以编程后， 文科生就可以统治世界了 一定要学编程！ 一定要学编程！ 一定要学编程！ 因为上帝就是个程序员 他只是不亲自写代码

@perkmaybe 我不信

Bro to Bro: build your x account Just say “hello” and gain 800 mutuals here.

@Jackywine I'm increasingly liking Codex, especially after Claude keeps getting worse.

我感觉我又可以了，Codex 太好用了你们知道吗？

@marmaduke091 Very looking forward to Image2, but even as a $200/month GPT Pro subscriber, I haven't received the grayscale test version yet.

🚨 GPT Image 2 is rolling out on ChatGPT now Check if you have it. You can test by prompting text heavy stuff. Bad news: They seem to have picked the WORST checkpoint, or maybe they have other plans. I don't want to lose my hope, let's just wait and see.

@Anas_founder I have a Mac mini m4 Pro, it only cost me 8000 RMB, it's simply the best product in the world

Developers, which one do you prefer? -Mac mini -MacBook Air