IBM Granite 3.0 Models are a series of high-performance AI language models developed by IBM and offered through the Ollama platform. These models are trained on over 12 trillion tokens, demonstrating significant improvements in performance and speed. They support tool-based use cases including retrieval-augmented generation (RAG), code generation, translation, and bug fixing. The IBM Granite 3.0 models include dense models and Mixture of Experts (MoE) models, the latter designed for low-latency use, suitable for on-device applications or scenarios requiring instant inference.