I'm curious about what models everyone has configured with Openclaw. Using Opus to run a website for two hours in the early morning completely exhausted the same-day transfer station quota.

Occasionally switching to the Sonnet model works pretty well for communication, but last week I bought the glm model and connected it to the API, and it felt very clunky with slow responses…
Early morning, I also bought Minimax 2.5, trying it out to help me finish the website I didn't complete yesterday. The prompts I give it are so rich that I could write a novel. Throughout, I kept saying "city gate tower," but it heard "hip bone axis."
In two hours, my Opus quota will refresh again, so I plan to add some more money to top it up.
I feel disembodied, and I’ve been crying.