The human microbiome plays a crucial role in maintaining our health. Alterations in the microbiome have been linked to various chronic conditions like autoimmune disorders, metabolic diseases, and cancer. While various tools have been developed to study the microbiome, each tool tends to be specialized for a specific task. This fragmented approach makes it difficult to gain a thorough picture of how the microbiome impacts health and disease. To overcome this limitation, here we report on the development of a foundation model—a model powered by generative AI—similar to the technology behind ChatGPT. While ChatGPT has been trained on massive text corpuses, our foundation model was pretrained on large-scale metagenomic sequencing data from the human gut microbiome, spanning over 13,000 microbiome samples. The model was then fine-tuned to predict host clinical status, distinguishing between healthy and disease states, as well as classifying 34 disease types. Our model demonstrated strong predictive performance, achieving an F1-score of 79.0% (AUC = 85.8%) in 10-fold cross-validation and an F1-score of 73.9% (AUC = 81.0%) on an external dataset for distinguishing healthy vs. diseased samples. Furthermore, it exhibited substantial capability in predicting disease type, with an average F1-score of 99.1% ±  1.5% (avg AUC = 98.2% ± 1.7%) across all 34 diseases in 10-fold cross-validation when predicting if a sample had that specific disease or was healthy. By providing a unified, scalable tool for microbiome analysis, this ongoing study has the potential to transform microbiome research and advance personalized medicine.