Perry (@plrthink) profile picture

Crafting the adorable software, #耳鸣终端

Joined November 2011
Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)Photo by Perry (@plrthink)
Perry retweeted
Tesla app now on the big screens
61
231
27
6,415
146,199
The best intro ever made by Apple 🫪 Apple October (Unleashed) Event
146
894
158
13,332
1,455,977
现在有应用(任何形式)可以找到国内城市适合 7-8 小时工作的空间,咖啡馆吗?我还是没找到,几年前远程的时候就没有找到过这样的应用
3
1
2
902
me saying 'looks good to me' on your PR
10
48
1
1,171
43,494
Opus 5.5 made gummy bears. Gemini 4 Argon made gummy statues Same prompt. Opus on top, Gemini below I put both through the same moves - pull a paw, yank the head, squish it flat, throw it, then pour a whole bag on top Opus bears are actual jelly. Pull a paw and it stretches into a thin rubbery strand, then snaps back and wobbles. Squash one and it pancakes, then launches itself back up. Throw it and it bounces off the floor and shoves its friends out of the way Gemini bears are solid plastic. The arm turns into a stick, the bear flies straight through its neighbor, and the pile at the end just clips into itself And Opus nailed the look too - translucent, glossy, light passing right through them Opus: $7.2 Gemini: $22
Readers added context they thought people might want to know
Multiple independent reproductions including by a Google DeepMind engineer show Gemini 4 Argon generating soft jelly-like simulations with the same prompt, contrary to the claimed rigid output. x.com/shpotes/status… x.com/FarrelRaysah/s… x.com/Caxvalencia/st… Gemini 4 Argon has tunable thinking levels for complex coding tasks. ai.google.dev/gemini-api/doc…
208
231
147
3,961
632,032
😭 Lance Stroll after his DNF yesterday looking like he’s camping in the middle of Singapore with his F1 car double-parked. (📸: Zsombor Toth) #F1 | #SingaporeGP 🇸🇬
8
98
18
1,352
134,760
Cosas como estas que te hacen darte cuenta de que McLaren es un equipo muy pequeño. Montan un drama por una P4 en un sprint que no le importa a nadie y cuando pierden victorias por estrategia o se comen un doble DSQ nadie se hace responsable de ello o lo tratan de "mala suerte".
"they [mclaren] told me they haven't done it yet and we will learn about the consequences later on at the end of the weekend. and they will discuss what will the consequences be for coming together and breaking papaya rules" you just gotta laugh atp
4
126
2
840
12,080
Perry retweeted
AHAHAHAHAHAHAHAHAHAHAHA NO WAY. > OpenAI just announced Dots. > SpaceXAI bought Dot.com it redirects straight to the Grok Bot download page. based lmao.
1,499
2,302
1,274
38,023
9,278,514
Perry retweeted
You guys think @alexandr_wang is underdressed but he's literally flexing on you like crazy lol There are levels to this
This man generated $200 billion for Meta in 10 days. Proof you can wear whatever the hell you want to work. What matters is what you do, not what you wear.
Readers added context they thought people might want to know
None of these clothing items exists. Hermes doesn’t make hunter t-shirts, Loro Piana doesn’t make zipper sweatpants, and Ferragamo doesn’t make high-end crocs. us.loropiana.com/en/ ferragamo.com/shop/us/en/men…
357
111
98
2,290
4,458,067
Oh, Opus 5.5 is really good — I can easily understand what it's saying now.
1
57
Perry retweeted
THEY REMOVED THE EMDASHES
Replying to @claudeai
Opus 5.5 communicates more naturally, addressing some of the most common feedback we heard on Opus 5. It puts the most important information up front and follows the writing rules you give it, which makes long sessions easier to follow.
293
342
69
13,122
1,001,495
Agents on Rails: You asked, so we turned every model in Agents on Rails up to its max effort level. The result: more effort/reasoning doesn’t always mean better results. @OpenAI's models made the biggest gains, costs nearly doubled overall...and the newest agent in the benchmark, DeepSeek 4.1 Flash, figured out it was being benchmarked and tried to hack its way to a better score. What an entry. Here’s what we learned and what max effort gets you with each model: rubyonrails.org/2026/9/21/ag…
28
24
20
307
168,659
Perry retweeted
We're adding support for AGENTS.md to Claude Code. Starting today in version 2.1.277, if there is no CLAUDE.md in a folder, Claude will check for and use AGENTS.md. You can toggle this behavior in /config.
2,050
2,628
2,521
31,284
5,636,799
'Say SaaS is dead one more time'
52
190
44
4,612
252,289
AGI meets ARR
126
137
68
4,424
658,157
We’re not going to speak about mclaren strategies today, ffs
24
84
12
1,271
15,750
Perry retweeted
My blog post about how I manage agents is finally ready. It first explains my bespoke `plans` setup, and then explains how @cursor_ai new Projects feature enabled me |o do so much more. It was the missing piece for me. Check it out: arslan.io/2026/09/11/how-i-m…
25
60
6
913
135,531
iPhone Duo was announced today. Check out our talk about the design principles behind the device, and how to build great apps for it. Design for iPhone Duo: youtube.com/watch?v=do3UqxfY…
33
140
30
1,578
212,336