>>105822298
The Claude models list endpoint (and the GPT one) don't have info about context size/max output tokens and Cohee NEEDS ST to cap those to the model value for you (he can't let the request just error and then display the error wtf that's insane).