TechPrufer
Get started
IBM
IBM logoIBM · Feature trail

IBM Granite Models

3 posts · Oct 2022 – Mar 2025

What this trail covers

IBM Granite models are IBM's family of open, performant, and trusted LLMs. This release includes Granite 3.0 language models for agentic workflows, RAG, text summarization, analytics, classification, and content generation. Granite for code models are trained on 116 programming languages for code-related tasks. Granite for time series models are for multivariate time-series forecasting. Granite Guardian models provide input-output guardrail capabilities. Models are available on Hugging Face, Ollama, and Replicate. The IBM Granite Community provides cookbooks and notebooks for data preparation, fine-tuning, RAG, tool use, and agentic workflows. The Granite Playground offers an interactive environment. A workshop guides users through common use cases. Integration with watsonx.ai Prompt Lab and watsonx Developer Hub is available.

All IBM trails
  1. Five ways to use the new Granite 3.2 modelsThis post details five use cases for the new Granite 3.2 models: turning PDFs into FAQs using multimodal RAG, debugging code from screenshots, building a personal stylist with vision and reasoning, creating an AI research agent for image analysis, and managing a budget from an image of spending history. It provides links to tutorials and demos for each use case.
  2. Discover the open, state-of-the-art IBM Granite modelsThis post announces the release of IBM Granite 3.0 language models, Granite for code models trained on 116 languages, Granite for time series models, and Granite Guardian guardrail models. It highlights their availability on Hugging Face, Ollama, and Replicate, and points to community resources like cookbooks, notebooks, and a playground for practical application. It also mentions integration with watsonx.ai and provides links to various tutorials and guided projects covering use cases like AI co-pilots, gen AI applications, fine-tuning, RAG, time series forecasting, and function calling.
  3. How to run AI inferencing on IBM Power10 leveraging MMAThis post details how to run AI inferencing on IBM Power10 processors by leveraging the new Matrix Math Accelerator (MMA) units. It explains the developer setup using Anaconda and Conda packages from the RocketCE channel, and presents a container-based demo framework with six real-world scenarios showcasing performance improvements (up to 3x) with MMA-optimized libraries compared to non-optimized ones. The demos cover image detection, action recognition, credit card fraud detection, and speech-to-text.