NoteLLM is a retrievable large language model designed for note recommendation, combining item-to-item (I2I) recommendation with hashtag and category generation. The model leverages large language models (LLMs) to enhance note embeddings by compressing note content into a single special token, which is then used for contrastive learning to identify related notes. Additionally, the model generates hashtags and categories through collaborative supervised fine-tuning (CSFT), improving note embeddings and recommendation quality. The framework includes a Note Compression Prompt to facilitate both I2I recommendation and generation tasks. Extensive experiments on real-world scenarios demonstrate that NoteLLM outperforms existing methods in note recommendation, particularly in handling low-exposure notes and improving recall and user engagement. The model also shows effectiveness in generating accurate hashtags and categories, enhancing the overall recommendation system. The integration of LLMs into note recommendation tasks offers significant improvements in both recommendation performance and user experience.NoteLLM is a retrievable large language model designed for note recommendation, combining item-to-item (I2I) recommendation with hashtag and category generation. The model leverages large language models (LLMs) to enhance note embeddings by compressing note content into a single special token, which is then used for contrastive learning to identify related notes. Additionally, the model generates hashtags and categories through collaborative supervised fine-tuning (CSFT), improving note embeddings and recommendation quality. The framework includes a Note Compression Prompt to facilitate both I2I recommendation and generation tasks. Extensive experiments on real-world scenarios demonstrate that NoteLLM outperforms existing methods in note recommendation, particularly in handling low-exposure notes and improving recall and user engagement. The model also shows effectiveness in generating accurate hashtags and categories, enhancing the overall recommendation system. The integration of LLMs into note recommendation tasks offers significant improvements in both recommendation performance and user experience.