Hacker News: conradkay

New comment by conradkay in "AI coding at home without going broke"

conradkay — Sat, 13 Jun 2026 22:35:41 +0000

I'm almost certain $2800 is actually too low if you're really hitting weekly limits.

I'm on the $100/m plan and used $300 at API billing yesterday (according to ccusage)

Seems like one session is >$100 and I can get 10 full sessions per week

The $200/m plan is supposed to be 4x that in usage, so with 2 of those you could use 4*2*100*10=$8000 in just a week

Using Simon's numbers here as a bare minimum https://simonwillison.net/2026/May/27/product-market-fit/#en... you'd get 1200*4*2=$9600 a month

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 22:25:15 +0000

If they ban GPUs we can always multiply the matrices on paper

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 22:23:06 +0000

Interesting comparison, thanks for sharing! It reminds me of this post about how machine learning and encryption have some fundamental similarities: https://reiner.org/neural-net-ciphers

> I can certainly imagine LLMs taking a similar path. Maybe it's useful to think about what fundamental differences could contribute to LLMs taking a very different path. What comes to mind is the scaling hypothesis, implying that the best LLMs will require enormous capital investment.

That seems largely incompatible with open source barring a fundamental change. There's open weights, but I can't think of a clean historical analogy there and find it extremely difficult to even guess how the future will go

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 21:49:59 +0000

My guess is that they liked the status quo with Project Glasswing and didn't want Fable to be public, especially if anyone is jailbreaking it into Mythos and using it for cyber

But then it backfired spectacularly and now it seems they can't use Mythos currently

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 21:35:09 +0000

I'd guess it's absurdly good business wise, given API pricing is like 10-30x as much

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 21:32:20 +0000

It's certainly worse news for Anthropic than other labs since it's not completely random, and there's people in the administration (e.g. David Sacks) who don't like Anthropic -- perhaps seeing them as an enemy

New comment by conradkay in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

conradkay — Sat, 13 Jun 2026 21:19:06 +0000

Seems like estimates are that 70-85% of their revenue comes from API usage/pricing, so some users switching from Opus to Fable for that would've had a big impact

Then there's people switching from GPT 5.5 or upgrading their subscriptions, and Fable being scheduled for removal from subscriptions on the 23rd

New comment by conradkay in "Claude Fable is relentlessly proactive"

conradkay — Fri, 12 Jun 2026 02:50:39 +0000

Does low/medium effort fix it for you? Seems like Fable 5 low can outperform Opus 4.8 high/xhigh often, and uses a lot fewer tokens

New comment by conradkay in "Claude Fable 5"

conradkay — Tue, 09 Jun 2026 17:32:38 +0000

Looking at page 255 of the model card (https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3...) it might be much better on all dimensions (speed, cost, quality) to just use Fable 5 on low/medium effort than switch to Opus

New comment by conradkay in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

conradkay — Tue, 02 Jun 2026 01:47:38 +0000

Looks like it'll be more like $2t

https://polymarket.com/event/anthropic-ipo-closing-market-ca...

New comment by conradkay in "Anthropic raises $65B in Series H funding at $965B post-money valuation"

conradkay — Fri, 29 May 2026 06:47:17 +0000

I suspect it's less true now that synthetic data has worked so well, and multimodal doesn't seem to transfer as well as many would've hoped

Claude Code and Codex are a big advantage, vs Gemini CLI which might be killedbygoogle soon? https://news.ycombinator.com/item?id=48196867

> Both Anthropic and OpenAI only has access to whatever they can buy or steal. A trillion can buy you quite a lot! Like offer some company a ton of money for data, and if they say no simply buy said company. Bonus points if it's someone like Atlassian who's stock price is getting hammered largely because of you.

New comment by conradkay in "Anthropic raises $65B in Series H funding at $965B post-money valuation"

conradkay — Fri, 29 May 2026 01:14:29 +0000

Yeah it's very murky, but if they're arguably profitable under some definition of profitable, then it's ridiculous to claim there's "no path to profitability"

Especially if you assume that 6 months ago they weren't very close to this version of profitable

New comment by conradkay in "Claude Opus 4.8"

conradkay — Thu, 28 May 2026 21:07:28 +0000

I think there's just less time between model releases now

New comment by conradkay in "Claude Opus 4.8"

conradkay — Thu, 28 May 2026 21:05:02 +0000

Probably not till mid June

New comment by conradkay in "Anthropic raises $65B in Series H funding at $965B post-money valuation"

conradkay — Thu, 28 May 2026 20:49:17 +0000

That feels really hard to argue now that Anthropic/OpenAI are so much bigger

How the hell do you crush a ~1T company on the one thing they have all their focus on?

New comment by conradkay in "Anthropic raises $65B in Series H funding at $965B post-money valuation"

conradkay — Thu, 28 May 2026 20:43:42 +0000

https://techcrunch.com/2026/05/20/anthropic-says-its-about-t...

New comment by conradkay in "Claude Code as a Daily Driver: Claude.md, Skills, Subagents, Plugins, and MCPs"

conradkay — Wed, 27 May 2026 23:30:56 +0000

I do find it somewhat concerning that the incentives aren't aligned, but as long as things live on your computer it should take just a prompt to migrate to something new

And Codex is open source

New comment by conradkay in "I think Anthropic and OpenAI have found product-market fit"

conradkay — Wed, 27 May 2026 20:36:40 +0000

GPT 5.5 is maybe 4x the size of v4 pro, hard to compare price since cache hit is basically free with Deepseek, but 40x cheaper (with their 75% off) seems about right.

So ballpark same price per parameter as Simon.

New comment by conradkay in "Project Glasswing: An Initial Update"

conradkay — Sat, 23 May 2026 04:06:24 +0000

You'll have access to the same models as your hypothetical attackers, and a big advantage if only you have access to the source code

New comment by conradkay in "Is AI Profitable Yet?"

conradkay — Sat, 23 May 2026 03:54:49 +0000

https://www.techinasia.com/news/chinas-deepseek-eyes-10b-fun...

Would be weird if they're raising $10 billion after spending only 0.3

https://newsletter.semianalysis.com/p/deepseek-debates

Probably more like 3-4 billion by now?