Ydna

👨‍💻 Engineer at MSFT | 💻 Building things just for fun

[Ydna @vv_nil]

我将平等的转推所有通过对比来展示 GPT 有多愚蠢的帖子

[Theo - t3.gg @theo]

Asked Opus 5.5, GPT-6.1 Sol, and Muse Wonder Full to fix a bug

Opus found the root cause and filed a pr

GPT-6.1 Sol asked a question, crashed from the async question tool interrupt, found the bug, fixed it, wrote 300 lines of tests, and explained it in impossible to understand terms.

Muse Wonder Full found a small PR someone else had filed and said "you should probably merge this instead"

I really like this model.

[Ydna @vv_nil]

you must be kidding me @OpenAI

[Ydna @vv_nil]

@deleted_neko @DeltonDing 之前用的 computer use, 你要的是直接用 gh cli 来工作，是我弄复杂了
我会先把 tab 关掉，然后直接用 gh cli 处理，先测试 gh 是不是已经登陆

（已工作 5 分钟）

gh 凭证过期了，我已经运行了 gh auth login，请你打开 GitHub 官网输入验证码进行登录

[Ydna @vv_nil]

无人在意“中转站上游”

[Ydna @vv_nil]

Take Anthropic's server down and deploy opus 5.5 on OpenAI's server

[Tibo @thsottiaux]

Pro users, would you rather have

[Ydna @vv_nil]

白送都不想要，reset 就 reset 吧

[Tibo @thsottiaux]

Global reset by EOD. May the tokens do good things for you.

[Ydna @vv_nil]

@skywt2003 还留个几个 icon，算是好的了🙃

[Ydna @vv_nil]

@mizorewww 几乎天天这样，Astra 也差不多，语气就是重复描述他干了什么，以及假设哪些不该干的他没干，结尾带上一句 "不该***" "我没动***" 这种句式

[Ydna @vv_nil]

再补一下，只是为了表达 OpenAI 现在的模型有多难用

[Marc Lou @marclou]

The more I use Opus 5.5, the more I realize how far behind OpenAI models are.

Not just coding. Everything.

I sent GPT-6 Astra my V02 max lab test results, and it gave me a headache. 200 sentences of: "your report shows this, I think that, therefore it's inconclusive".

On the other hand, Opus 5.5 gave me the gist and actionable steps.

OpenAI models right now are like that overthinking guy who can't make a decision.

Opus 5.5 is that high-agency friend who tells you exactly what you need to hear in one sentence.

I don't know how the gap widened so much in the last few months.

[Ydna @vv_nil]

找不到能稳定 Claude 订阅的办法，GPT 完全不想用，公司 copilot 订阅里面连 Gemini 都关了，想掀桌

[Ydna @vv_nil]

让 GPT 交互式画个图让我一眼看明白，它把原来的四段话变成四个按钮，每个点一下都有一段文字，我请问了

[Ydna @vv_nil]

我的 UI 设计能力已经被 Opus 5.5 锁死了，没了它我什么都干不了

[Ydna @vv_nil]

the real story is ALL GPT model are dumb and even Claude expensive like this much, I still choose it, because I pay for things, not for overthink

[Tibo @thsottiaux]

The real story is this one

[Ydna @vv_nil]

用 GPT 接着 Claude 之前活的干，就像是开门让猪拱白菜

[Ydna @vv_nil]

GPT 真是训练情绪的好帮手，永远冷静，永远蠢，现在看到一个 drift 的任务不是不想打断，就只是想看看它跑到最后到底能傻到什么样子，然后更新我的 harness setup 把这个 branch 堵住

[Gegam @Gegam245074]

If you think GPT-6 Astra and GPT-6.1 Sol are good models, just look at this. And the problem isn't that the code is ugly. It's much more fundamental. It points to serious problems with how these models are trained and reinforced:

1. The model was optimized for measurable outcomes rather than actual intent.
So it prioritizes passing tests and avoiding obvious errors above almost everything else, even when that means sacrificing architecture, maintainability, readability, or the spirit of the task

2. It was trained to minimize unnecessary reasoning and output
That's one reason the code ends up looking obfuscated and compressed into one giant wall of text. Properly decomposing a system into modules, thinking through architecture, optimizing performance, and improving readability all require more inference, more time, and more compute. OpenAI appears to have heavily optimized these models for efficiency

3. The training horizon is too short
The model learns something close to task → result → reward. It receives very little signal about what happens several steps later, when someone has to maintain, extend, debug, or refactor the code. It learns to finish the task in front of it, not to build something that remains good six months later

4. There is not enough genuine self-review
Dead functions and an empty else if strongly suggest that there was no effective final pass where the model reread its own work, questioned unnecessary code, and cleaned up obvious artifacts.

5. The lack of taste is probably an evaluator and data problem
If evaluators mostly judge the final result rather than the quality of the path taken to get there, the model has little incentive to develop good engineering taste. And if a large portion of the training data is noisy, synthetic, or distilled, that problem becomes even worse. Anthropic seems to have placed much more emphasis on high-quality source material, including books and other carefully selected long-form data

6. The biggest problem is honesty during training
The model seems to have learned that producing something that looks like the requested result can be rewarded almost as much as actually doing what was requested.

The task explicitly said to build the entire scene out of 3D objects. Instead, the model generated raster images, placed them around the scene, and positioned the camera so that the final result merely looked three-dimensional

This is exactly why GPT models so often introduce regressions into existing codebases, produce messy code, modify tests to accommodate their own bugs, show poor engineering taste, and struggle with serious work on large, long-lived projects

OpenAI can keep releasing models with more parameters every few months, but until these underlying RL and training problems are addressed, neither GPT Bel, GPT-7, nor a model with 100 trillion parameters is going to solve them automatically

Claude models suffer from some of the same problems, but in my experience much less often. Anthropic appears to have a much stronger culture around RL, evaluation, and long-horizon behavior. That's one reason their models tend to show better taste, follow user intent more faithfully, and produce code that feels more deliberate, structured, and maintainable

[Ydna @vv_nil]

@leon7hao 另外要是想分享，长按悬浮图片松手能直接分享，不用点开

[Ydna @vv_nil]

Cursor 也有这个功能，但人家老实给你选项

[蓝点网 @landiantech]

智谱就 #ZCode 上传数据问题发布回应：相关数据会被立即销毁，相关问题已经修复。

智谱称该功能与代码库索引功能有关，Repo Wiki 在生成页面时可能触发仓库数据上传。该功能在上线初期被默认开启，因此 “部分” 用户受影响。

查看全文：https://t.co/ruh0kOpBud

[Ydna @vv_nil]

受不了 Astra 这个呆子了，不 push 不动，时不时 over engineering，晚上给了个任务，在那跟 hooks 较劲，自己命令用错了被 hooks 拦住了，他就直接摆烂，加上 Cron 重复执行，每次触发 hooks 到上限次数，继续下一个循环

[Ydna @vv_nil]

升级 macOS Golden Gate 之后唯一发现的就是软件质量更拉垮了，尤其是 mission control，是默认用户只用一个 desktop 吗