hang (@think_tru) profile picture

build KeyPaste: https://teapawt.net/t.co/T2fXGv3VhQ SimpleSpend: https://teapawt.net/t.co/fBe8yWnlU9

Joined November 2015
Photo by hang (@think_tru)Photo by hang (@think_tru)
在grok网页对话上一直使用Heavy模式,没想到额度消耗量还是很大的
55
用了一段时间Grok build,语音输入比Codex好用多了,整体感觉更快更准,TUI也能支持语音输入,有点意外
1
69
codex的侧栏模拟器映射竟然能支持双指滑动页面,这下要多用这个功能了
44
简单用了一下新出的ChatGPT Voice,总体感觉还不是太可用。首先问题就是他回答的内容太长,信息密度低,效率不高。不过未来可期,继续等吧
42
想试试Grok了
We've doubled usage limits for all individual and teams plans! These limits apply to Grok, Composer, and any new Cursor models.
46
M5 Pro,24G内存的配置,在这个agent时代已经不够用了 Xcode运行iOS测试,自动启动5个模拟器之后,内存已经告急,有时候需要开一个macOS虚拟机跑macOS的测试,就需要考虑系统资源的问题。 Xcode cloud的额度也不够,现在看来下一步应该配置一个Mac mini来缓解主力Mac的工作
115
At this point, Chrome on my Mac is basically just for AI. I use Safari myself and let my agents do their work in Chrome. 🤤
33
I just noticed that Grok 4.5 is now available in Cursor and shares the same pooled usage allowance as Composer 2.5. Cursor still feels very competitive right now.
1
63
I’ve subscribed to Cursor twice, and the overall experience has been great. Composer 2.5 offers plenty of usage, and Fast Mode is incredibly quick. Grok 4.5 itself is also fast. I’ll probably resubscribe when I get the chance and see how well it performs.
66
After using GPT-5.6, I feel that agents have become much better at computer use. Manual acceptance testing is taking too much time—it may be time to hand more of that work over to AI.
43
最近更新工作流程以及新的模型GPT-5.6用上之后,出现了两个问题。 一个是Token不够用了,现在两个小时就耗尽codex 5小时的额度,一天能跑掉50%的周额度(100的套餐) 另一个是功能验收太耗时了,需求完成的速度远大于手动验收的速度,完成的需求都堆积在PR上。
81
今天使用了一天 GPT-5.6-sol,最大的感受是,额度用的更快了,之前差不多的工作流程,三四个小时才回耗尽5小时额度,今天基本一两个小时就快到底了。 不过用 sol 解决了几个之前 5.5 xhigh 都没有解决的问题,能力上确实明显更强了。
1
89
Assuming OpenAI is already using GPT-5.6, or even more powerful models, internally to develop their products, does that mean even stronger models still can’t fix this ChatGPT bug on macOS? I’ve seen this issue multiple times already.
1
1
432