IBM Launches Granite 4.2 Models for Local and Agentic Enterprise AI
IBM has launched its new Granite 4.2 language models, designed specifically for local execution. These models focus on agentic capabilities and predictable enterprise deployment. This release addresses the growing demand for local LLM deployment, allowing enterprises to run AI models on-premise for better data privacy and control. The emphasis on agentic capabilities also supports the shift toward autonomous AI workflows that can execute multi-step tasks. The Granite 4.2 models are optimized for local execution and enterprise environments, prioritizing predictable performance and agentic workflows. They build on IBM's open-weights approach, allowing customization and deployment across various business use cases.
## BACKGROUND
IBM's Granite is a family of open, trusted AI models designed for business applications, covering language, vision, and speech. Agentic LLMs refer to models capable of acting autonomously, using tools, and executing multi-step tasks to achieve specific goals rather than just generating text.