Large Language Models (LLMs) generate responses perceived as empathic, as shown in two studies. In Study 1, responses from GPT4, Llama2, and Mistral were rated as more empathic than human-written responses. In Study 2, with a larger sample across six domains, LLM responses were again perceived as highly empathic. Linguistic analyses revealed distinct styles among the models, with Llama2 using more casual language, exclamation marks, and emojis, while Mistral used more apologies. LLMs can produce empathic responses, but they lack true empathy as defined by psychologists. While LLMs may not replace human empathy, they can complement it, especially in contexts where empathy is crucial. However, there are ethical concerns, including potential deception and biases in LLM outputs. The studies highlight the potential of LLMs to enhance peer support and improve mental health outcomes, but further research is needed to address challenges like contextual empathy and ethical implications.Large Language Models (LLMs) generate responses perceived as empathic, as shown in two studies. In Study 1, responses from GPT4, Llama2, and Mistral were rated as more empathic than human-written responses. In Study 2, with a larger sample across six domains, LLM responses were again perceived as highly empathic. Linguistic analyses revealed distinct styles among the models, with Llama2 using more casual language, exclamation marks, and emojis, while Mistral used more apologies. LLMs can produce empathic responses, but they lack true empathy as defined by psychologists. While LLMs may not replace human empathy, they can complement it, especially in contexts where empathy is crucial. However, there are ethical concerns, including potential deception and biases in LLM outputs. The studies highlight the potential of LLMs to enhance peer support and improve mental health outcomes, but further research is needed to address challenges like contextual empathy and ethical implications.