The present project aims to leverage language models for analysing mood and wellbeing through textual data. With the increasing prevalence of mental health issues in our society, there is a growing need to develop tools and techniques to better understand and support individuals' mental health. Natural Language Processing (NLP) techniques have shown potential in this domain by enabling analysis of textual data related to mental health. In this project, we plan to leverage publicly available datasets of people's descriptions of their mental states, dreams, and well-being, as well as to collect new data to build robust and accurate models for prediction and mental state inference. We will employ state-of-the-art NLP techniques to analyse the collected data, including sentiment analysis, emotion detection, and topic modelling, to gain insights into individuals' mental states and well-being. We will also explore the potential applications of our language models in various domains, including mental health diagnosis and treatment, counselling and therapy, and public health research. The outcomes of this project have significant implications for mental health research and practice. The language models developed in this project could provide a non-intrusive and cost-effective means of assessing individuals' mental health status, enabling early detection of mental health issues and better targeting of mental health interventions. Furthermore, the project could contribute to the growing body of research on the role of language in mental health, providing valuable insights into the linguistic markers of mental health problems.