Inception-v3 has about 50 layers, and that's considered a lot (requires considerable processing power to train).