IBM's Granite 4.2 Brings AI Reasoning to the Enterprise

IBM has released Granite 4.2, an enterprise-focused AI model designed to make reasoning cheap, deployable, and reliable at scale. Unlike consumer chatbots, Granite 4.2 is aimed at business use cases, where companies need AI agents that can automate tasks, write code, and run terminal commands.

The model comes in three sizes: 3B, 8B, and 30B, each with different capabilities. The smaller 3B model can run on a developer's laptop, while the larger 8B and 30B models require more powerful hardware. The 8B and 30B models received additional training in coding, terminal, and search environments to learn agent behaviors.

Granite 4.2 is built using IBM's CodeAlchemy pipeline and trained on roughly 15 trillion tokens. It has a compact model family designed to stay deployable, compliant, and cost-efficient. The result is a set of AI models that can perform reasoning tasks while staying within budget.