• 0 Posts
  • 89 Comments
Joined 1 month ago
cake
Cake day: August 12th, 2026

help-circle











  • I prefer it less for the writing part (unless big refactors that aren’t actually complex) and more for analysis. If you got a couple million lines of code upstream that you depend on heavily (ERP modules) then it’s nice to fire off a prompt and come back to “you got weird behavior x because upstream changed y, here’s the exact file and lines and how it flows there”








  • If you think about it, with open weight LLMs… Concentration of power is reversed if anything.

    Used to be if you didn’t know how to program and you needed a super specific tool for your own use case, you better have tons of money. Need artwork for your startup on an already tight budget? Pay up.

    Things that were accessible to people with either skills or money, are now accessible to people with a few dollars to throw at GLM or Qwen or a beefy enough machine to run a good model at home. Still not free, but much cheaper than contacting someone.

    My Claude subscription ran out and I ran GLM 5.3 flash to do a review on my gigantic PR (massive refactoring to keep up with upstream changes, not adding a ton of vibe coded “features”) and it gave me a pretty good analysis of the weak spots. It costs 1/100 of Claude Fable or 1/50 of Claude Opus to run, API pricing to API pricing. More amazed with this than anything Qwen right now. Beat Deepseek V4 Flash which I was really fond of for a while due to its price to performance ratio.