Comment by johndoe18637

Could you check how many tokens the different models spent to do the task? With all the comments about Kimi K3 Not being token efficient I'm curious if your test confirms that