Matt Rubens

Grok 4.5 is a great model, and also the only frontier model that makes typos. Almost too human? Seriously though, I wonder why.

@mitchellh Franklin from GPT 5

Spent all day exclusively using Sol Ultra to get a feel for it. No notes yet still figuring it all out. But this caught me off guard. Like who the fuck is Franklin. lol. (Obv I clicked and found out but when I tabbed over I was taken aback)

@creatine_cycle From accounting?

having a gf is insane because it's literally sleepy claude

@CtrlAltDwayne I need Fable to parse this sentence

Seen a lot of people saying that Fable 5 can't be used for coding, which is not what Anthropic said at all. Some of you have skill issues around reading comprehension. It says "some" not all "some" meaning the classifier is probably sensitive depending on the task.

@udiWertheimer Fable for parsing your screenshots, GLM for boring stuff like coding and debugging

watch as nobody mentions GLM 5.2 ever again

@mattshumer_ My bugs are a threat to national security

So uh, Fable will be useless? Apparently coding work will fall back to Opus.

@signulll Awaiting the graphic that shows which coding/debugging tasks go to Fable and which go to Opus

whoever is writing this stuff on the anthropic blog & making these graphics deserves kudos. the posts are genuinely fun to read, but also informative, crisp, clean, & clear enough for normal peeps to understand without necessarily being condescended to. that is an extremely underrated skill. making complex systems more relatable without making ppl feel stupid is hard af. i secretly hope it is fable or mythos writing them & also producing the graphics.

So this is how it feels to be part of the permanent underclass

Mythos please build GTA6 make no mistakes

@jxnlco I have yet to see anyone actually do less work because of AI

loop this loop that but honestly, if you get good enough at using codex with a orchestration loop, you too can be one of those people at equinox at 11:20am on a tuesday morning. "make up the chief of staff thread and then every 100 minutes, check all my connectors coordinate all the work across my pinned threads"

@arjundundoo Andrew tell me what to do

as you may have guessed, I took the offer at Bird kms

@MoritzW42 Mythos would never

holy shit - their api is leaking customer data

Alert TBPN

@jxnlco @wonforall Kobe would never pass control to a subagent

insane ball knowledge in codex I just found out @wonforall has a skill called $kobe that spawns off 3 subagents to discuss / review his code, each of which is build to represent one of our principal engineers on tuned in on his past code reviews. I'm going to start doing this with @dkundel and @charlierguo for our docs...

@bhalligan Why didn't the engineer learn from the session?

When your engineer finishes a session with a coding agent, they have more code than they started with and exactly the same amount of skill. The agent produced output. It did not produce understanding. You paid frontier prices to make your codebase bigger and your engineer no smarter. That is not an accident. It is a design choice, and it is worth remebering who it serves.

@garrytan Inside each of us there are two wolves.

There are two loops in every founder's head. The autism loop: run your own model to the floor, ignore consensus, hold a thesis when everyone says you're wrong. That makes conviction. The empathy loop: feel what the user feels, sense what the market wants before it has words. That makes traction. Most people crank one and starve the other. Pure conviction builds something brilliant nobody wants. Pure empathy builds consensus mush. PG put the whole job in four words: make something people want. The autism loop makes the something. The empathy loop knows it's wanted. The founder is the bridge. Most great founders show up dominant in the first loop. That's why they're contrarian enough to try at all. The work is grafting on the second. There is no place in the world that helps founders make the two loops work together to make great startups than Y Combinator. It is the most gratifying part of our work.

@RhysSullivan @Jason Beg to differ when it comes to coding harnesses at least

@Jason ...as opposed to if they just paid for Anthropic / OpenAI tokens then the labs wouldn't do that? that's provably false the labs: 1) don't ship high quality products 2) have too little focus if your product can lose to them copying you then you were ngmi to begin with

Fair warning, YC founders: if you take these tokens, there’s a non-zero chance that OpenAI will study exactly what your startup is doing, copy your idea and put your app into their free offering. This is the classic platform playbook — be careful, founders!

@sama How much equity for 6 weeks of @steipete token usage?

i am excited to see what will happen with tokenmaxxing startups, both for how they work internally and the products they can build. openai offered to invest $2M in tokens into every startup in the current yc batch. happy building!

@asdfgaskldjgna Feel free to keep using it if it works for you!

@mattrubens gpt pro 구독 끝나고 오늘 처음으로 로컬llm으로 써보는데 느리지만 codex 5.5 very high랑 비빌정도로 좋은데 아니 내일 문닫는다니!!!!!!ㅠㅠㅠㅠㅠㅠㅠ

@thorstenball The coding agent is undead!

The best coding agent in the world is back. ampcode.com/news/neo