justin
@justinsunyt
Followers
6K
Following
19K
Media
925
Statuses
5K
helping @capydotai code itself • prev @pennmandt
sf
Joined May 2023
we spent $100k on the cutest domain in tech and shipped a brand new way to code with ai
106
37
680
peak is back
ෆ.*･ﾟ┈┈┈┈┈┈┈┈┈.*･ﾟෆ かぐや様は告らせたい 大人への階段 本PV ෆ.*･ﾟ┈┈┈┈┈┈┈┈┈ .*･ﾟෆ ♪OPテーマ 鈴木雅之 feat. 古賀 葵「アブナイキオク」 https://t.co/7Tg0IdT5pp ᥫᩣ𝟐𝟎𝟐𝟓.𝟏𝟐.𝟑𝟏 𝟐𝟐:𝟎𝟎 𝐎𝐍𝐀𝐈𝐑 https://t.co/DPsdIaBcGS #かぐや様
3
0
4
why are the new york airports so goddamn far away from manhattan sf wins this one
2
0
7
Claude the Alligator's Death
2
0
16
Submit your CGT research by January 30 for the ASGCT 2026 Annual Meeting, held in Boston on May 11-15!
0
19
91
how are people even mad at ai summaries
spotify wrapped using AI again this year sigh
3
0
42
Without naming your job, tell me something you say 20 times per day at work
12
2
173
what the fuck
Anthropic is acquiring @bunjavascript to further accelerate Claude Code’s growth. We're delighted that Bun—which has dramatically improved the JavaScript and TypeScript developer experience—is joining us to make Claude Code even better. Read more:
0
0
54
opus 4.5 on capy is nuts i just vibe coded a whole expo app from scratch - with just a few tasks the whole app is 100% functional and looking amazing
12
1
220
finally, a proper reasoning selector for every model say goodbye to configuring 15 versions of gpt-5.1 for your coding agent
finally, a model selector with pricing, effective context length, and estimated tps
29
45
870
never hire a shitposter
2
0
45
“code on your phone without terminal emulator, tmux, and ssh” challenge impossible
3
0
26
LLM APIs are becoming less and less unified with every model release like wdym there are 7 reasoning efforts we have to support in our coding agent
2
0
12
claude “thinking” models have always done this since their reasoning is just fancy cot - without thinking they start to reason out loud in assistant text after a few turns if you want clean chat/file outputs at the expense of a few extra tokens always leave thinking on
I looked into this and the answer is so funny. In the No Thinking setting, Opus 4.5 repurposes the Python tool to have an extended chain of thought. It just writes long comments, prints something simple, and loops! Here's how it starts one problem:
0
0
8