Remember, just these three are enough.
First place: Claude Code
Second place: Codex
Third place: Grok
I’ve seen some products developed with domestic AI models, and they’re honestly unbearable to look at—they’re just not on the same level as Claude.
Seriously, if you want to develop a product that can actually make money, stop using GLM and Kimi and the like.
when you share api key again? im waiting for you
There’s no doubt it’s CC.
I often wonder if GPT lacks perspective, like a stubborn old ox.
Just use whichever one you can get for free 
Detail 3 — I actually fell into that trap myself. It took me two days to get out of it! If only I’d seen your post earlier. That’s all from me, I’ll go back to lurking now.
If you can use it for free, it’s a good API.
It’s all free stuff.
Okay, okay.
Thank you for the recommendation.
Thanks for the recommendation, I’ll give it a try.
I personally think GLM and Kimi are pretty good. Maybe you got routed to Doubao (just kidding).
Kimi and GLM are still quite good. GPT and CC are sometimes simply terrible at Chinese, so they can complement each other.
glm only writes bugs. Try to avoid it if possible.
I generally use Codex for auditing; Chinese models are actually pretty decent after a few rounds of revisions.
Is the saying “free things are the most expensive” still on the rise? But I don’t have any money!
My workplace requires me to use Claude Code, but I’m already retired and don’t really do any coding anymore. I haven’t used any of these tools—mostly just write and polish simple documents, and I think they’re probably all like Doubao-type tools.