Salesforce has launched Koa, its first specialized reasoning model built on Nvidia’s Nemotron-3-Super-120B foundation model. Designed for the Agentforce platform, the model targets sales pipelines, marketing campaigns, and customer support workflows. Koa employs a reinforcement learning technique called GRPO and features a 1-million-token context window. It outperforms the base Nemotron model it was derived from, particularly in multi-turn tool-use scenarios. The Salesforce-Nvidia partnership, announced in March 2026, also emphasizes governance and compliance capabilities for regulated industries like financial services and healthcare.
Source: Read the original article
