Are DeepSeek R1 And Other Reasoning Models More Faithful?