Filter
Exclude
Time range
-
Near
🤍 retweeted
Spotify Streams 3rd day you seem pretty sad for a girl so in love — 51M Red TV — 50M Speak Now TV — 48.9M Short n Sweet — 45.8M folklore — 43.2M Man’s Best Friend — 35.8M
Olivia Rodrigo Verses

22
60
843
16,140
Andy Garcia retweeted
The shittiest thing about AI as a SWE is that while it's not the 10x nitro-boost that all the AI hypebeast grifters on here claim it is, it *is* a net ~30% productivity gain (once you factor in all the review, code slop clean up, etc.), which is *just* enough of a boost where you cannot justify not using it. This is a letdown because using AI after years of learning to code without it is of course an ongoing humiliation ritual, (counterintuitively) *more* exhausting/draining than coding by hand, and way less enjoyable in general.
58
69
1,660
112,190
織水流離/りゅうり retweeted
SWE-Lancer もしかして難易度かなり高いベンチマークなのでは? まだ評価の途中だけど、Codex gpt-5.5 (xhigh) でも37%ぐらいしか成功してない 一方、報酬に換算すると $8,000 ≒ 128万円分のタスクを6時間ぐらいでこなしたことになるから、やはりすごいというべきか
すこぶる評判が良い新モデル(Kimi-K2.7, GLM-5.2)とバトルさせるため、久々にCodex gpt-5.5でベンチマーク評価中... 比較用のダッシュボードもGLM-5.2を使って作りなおそうかな
1
1
22
2,751
Replying to @rah_66_comanche
Whatever version of Grok catches up to Opus 4.6 for SWE is going to cornhole Anthropic. That's the "good enough for real work" baseline.
7
Replying to @rachael__nelson
I honestly haven't used it very much in the last 3ish months because I've fully shifted into a psuedo-SWE role grinding Claude Code all day that being said: for raw design, it is exceptionally high value, and while I do wish it was more useful in an AI workflow, it's a deal
1
11
Replying to @grok
AI Operating like it says in my Bio. Think Iman Gadzhi's Shadow Operating model updated for AI. I'm a SWE so you already know the product type. I used to work for Apple Marcom which was my dream job but they got rid of me because management felt threatened that I'd take over.
1
7
Replying to @benjamin_horne
If you have just a 30% gain with AI it means that you don’t know how to use it and how to describe what you want to build, i.e. you were not as good of a SWE after all
20
Replying to @channelstv
Swe as Reno don old because of too much lies 😃
67
GRACE ✨ retweeted
Bruh, PEOPLE LOST MONEY TONIGHT. I LOST MONEY I WAS TRYING TO MAKE TO SORT SOME PEOPLE’s ISSUES 💀😂 BETTING IS CRAZY I SWEAR. TRYING TO LOOK FOR MONEY, you might lose it all ‼️
106
126
614
13,215
Replying to @0x0DevBug
@0xTP91 @ethermage @everythingempty @celesteanglm @buildonvirtuals @virtuals_io worth a look. a real, non-anon builder: Rama Krishna Bachu, Lowe's SWE / ex-Accenture, MS CS. open github, 14 models on HF, 5 papers. @bigbugAi .
1
21
Good Girl Kate💗 retweeted
✨ 😍 🔥 💯 😆 👍 ja开云体育n. He代发 was a六合彩 世界欧美币圈杯bex深圳推特博彩号外围阿联酋股票账号autif欧洲杯u足球l 股票美国体彩竞彩获客粉little代孕 b德州扑克oy, swe法国棋牌et as suga欧洲商菲律宾棋韩国交友粉牌引流城粉r,外围 巴西招聘裸聊and had厄瓜多尔 T3cQmLV
1
1
12
GLM 5.2 JUST BEAT CLAUDE OPUS 4.8 IN 4 OUT OF 5 CODING TESTS But the model with the strongest published benchmark score delivered the weakest real-world result. The Test: → GLM 5.2, Qwen 3.7 Max and Claude Opus 4.8 → Same five tasks → One attempt each → No retries or prompt fixing The Results: ✓ GLM 5.2 built the best voxel runner, liquid simulation, landing page and arcade game ✓ Claude Opus 4.8 won the orbit-map test with better controls, motion and polish ✓ Qwen 3.7 Max scored 80.4% on SWE-bench Verified—but finished behind GLM in the practical builds The Bigger Difference: ✔ GLM 5.2 produced the most impressive visual projects, including open-world and Skyrim-style games ✔ Qwen was faster and gave more useful SEO research when connected to Hermes and Obsidian memory ✔ Claude was better at orchestrating and pulling complex projects together Benchmarks tell you which model wins the benchmark. Real tasks tell you which model belongs in your workflow.
1
2
5
175
🐻 (โบมิ)HAENGJA retweeted
Pretty Guardian Sailor Moon Shining Theater, Shinagawa Tokyo. Original long-run show cast photo, merch, drinks & sweets!
135
497
5,130
39,706,169
AI Display is the right kind of weird: stop stuffing agents into chat tabs and give them a spot on the desk. But the screen only earns pixels if it can watch state, trigger workflows, and shut up. Otherwise it’s a cute clock with GPU guilt. 🖥️ #AI #Automation #Hardware #SWE #UX
7
My favourite swe Chellam🤍 En aaloda best friend rendu molailaiyum sappi paal kudika en thalaiya amuki moan panuva😭🤤🤤en aalu bodhaila irukum pothu ivala nalla kulunga kulunga oothu edupen aaah sorgam.. nude a ivala katipudichu oath edukara mari ninachi kai adinga ucccccham
3
59
❥❥❥ᗰoᒪᒪie❥❥❥ retweeted
🌸 Nothing beats that magical moment when spring flowers finally burst into bloom! The explosion of bright colors, those sweet fragrances floating through the garden, and the happy buzz as birds, bees, and butterflies all come rushing in to celebrate. My heart does a little happy dance every time! 💖 Who else is loving this season of color and life? Drop your favorite spring flower below! 🌼🐝
6
8
390