Why nobody talks here about how many mistakes and how unreliable are the models? All, with no exception. Astra at Ultra included? Try to give a task to any of them, any medium complexity task. Then ask them “are you 100% sure?”. You will see that also 100%…