Im cancelling my Claude Code subscription | Hacker News Reader

Im cancelling my Claude Code subscription

numerlab.org

8 pointswallach-game10 comments

Performance and costs aside, the fact that they suspend your account with Zero reasons given, zero care for the appeal process, and don’t even have a broken archive download process should make everyone question whether they deserve their money or can be a reliable provider or not.

https://news.ycombinator.com/item?id=48597861

rofl rn

Ill tested deepseek v4 flash vs claude sonent and opus, and the results amazed me, if you wanna findout more checkout my article.

i checked out the article and while i find interesting the topic in general i think the blog post should be optimized in terms of writing, the mix of lowercase "i" and em dashes gives a bad impression you knowwhy, together with the writing style.

an improvement could be more structured and direct text, more structured comparisons (tables) and less story like, think of a typical technical articles focusing on relevant details and presenting them neutrally and structurally

i will try better next time, thanks for your input

I wonder how long it would take a "normal" coding prompt to go thru a "1T-2T" models on a medium performance consumer desktop.

Hours? Days?

https://www.youtube.com/watch?v=9TyJ9s26ylc

acoording to this video, around 30t/s, if in 2027 some 256gb ai box arrive, the performance will be even better

I'll watch this video once I am on my yt-dlp rig. Hopefully, those are 1T-2T params models.

If "AI" becomes ubiquitous, it may end up pushed out of the 'standard economy' and made sorta an 'utility' (to speak using US words). Because, you will have those able to pay for performant AI... and the other ones. To mitigate that, running locally at decent (but still slow) speed, frontier models, may become critical.

I say that: I have never fooled around with AI since they are gated with 'whatwg cartel' web engines.

I did watch your vid: it is for quantized deepseek models. I would look for full weight inference for maximum quality.

For instance, do we have similar benchmarks on the latest frontier open weight KIMI model at several tera-weights (I would be interested at coding prompts)?

okay but what was code quality of each?

Reply on news.ycombinator.com