Hi Yohan, thank you for your patience. After doing some digging, we found a setting that was incorrectly giving your models the wrong computation time. We have manually switch back the max. computation time of the 4 models you shared above and reset their cache.
Could you try using them again on your platform?