Recent advancements in artificial intelligence (AI), particularly large language models (LLMs) and multimodal foundation models, are transforming precision oncology. Since 2022, AI has significantly accelerated, with LLMs now achieving human-level text processing. These models, based on transformer architectures, enable multimodal AI systems that integrate diverse data types, such as text, images, and numerical data, marking a shift from specialized models. The integration of AI in oncology has evolved from early 2012, when CNNs revolutionized image processing, to the current era of LLMs and multimodal models. LLMs process and generate text-based data, trained on vast amounts of text, including medical data. They show promise in healthcare, particularly in medical text mining and clinical outcome prediction. Models like Bio-BERT and Med-PaLM have demonstrated success in medical applications. Additionally, "Retrieval Augmented Generation" (RAG) allows LLMs to use domain-specific knowledge for medical tasks. Multimodal AI systems can interpret multiple data types, such as text and images, and are being evaluated for precision oncology applications, including outcome prediction. However, challenges remain, including data quality, model interpretability, and regulatory approval. Foundation models, pre-trained on diverse tasks, reduce the need for specialized data, improving performance in disease detection and drug discovery. Despite progress, challenges such as model interpretability and regulatory hurdles must be addressed to fully realize the potential of AI in oncology. This editorial highlights the transformative potential of LLMs and multimodal models in precision oncology, emphasizing the need for interdisciplinary collaboration and rigorous validation.Recent advancements in artificial intelligence (AI), particularly large language models (LLMs) and multimodal foundation models, are transforming precision oncology. Since 2022, AI has significantly accelerated, with LLMs now achieving human-level text processing. These models, based on transformer architectures, enable multimodal AI systems that integrate diverse data types, such as text, images, and numerical data, marking a shift from specialized models. The integration of AI in oncology has evolved from early 2012, when CNNs revolutionized image processing, to the current era of LLMs and multimodal models. LLMs process and generate text-based data, trained on vast amounts of text, including medical data. They show promise in healthcare, particularly in medical text mining and clinical outcome prediction. Models like Bio-BERT and Med-PaLM have demonstrated success in medical applications. Additionally, "Retrieval Augmented Generation" (RAG) allows LLMs to use domain-specific knowledge for medical tasks. Multimodal AI systems can interpret multiple data types, such as text and images, and are being evaluated for precision oncology applications, including outcome prediction. However, challenges remain, including data quality, model interpretability, and regulatory approval. Foundation models, pre-trained on diverse tasks, reduce the need for specialized data, improving performance in disease detection and drug discovery. Despite progress, challenges such as model interpretability and regulatory hurdles must be addressed to fully realize the potential of AI in oncology. This editorial highlights the transformative potential of LLMs and multimodal models in precision oncology, emphasizing the need for interdisciplinary collaboration and rigorous validation.