IBM Launches Granite 4.2 Open-Weight LLMs for Local Deployment

IBM has introduced Granite 4.2, its newest family of open-weight large language models designed to be downloaded and self-hosted, available in 3B, 8B, and 30B parameter sizes. All variants use a decoder-only architecture and offer a native 128,000-token context window. The 8B and 30B models received additional agentic reinforcement learning to improve capabilities such as terminal use, web search, and external tool access, while the 3B model also supports tools but without the same level of specialized training.