Nvidia and Servicenow have collaborated in the development of a model Large language (LLM) open source that will allow the creation and development of AI agents. The model, presented at the Knowledge 2025 annual conference of Servicenow, is called Apriel Nemotron 15Band it has been created with the objective of offering low latency and equally reduced inference costs in the work with agriculture to automate company workloads.
According to its creators, it has trained using NVIDIA NEMO, as well as the Nvidia data set Llama Nemotron Post-Training. In addition, in the process they have also used internal Servicenow data.
Among its main advantages is the integration of advanced reasoning capabilities in a size smaller than that of other models, since it has only 15,000 million parameters compared to the three billion other LLMS. Thanks to this, it is a model whose training is faster and more economical when done using NVIDIA GPU infrastructure as a NVIDIA NIM microservice, also offering business level intelligence looking for organizations.
Apparently, nemotron 15b has shown promising results for its model category in the tests that have been performed, which confirms its usefulness for the development of agricultural AI and workflows related to it.
Apart will be available throughout the second quarter of this year, and will facilitate the creation of AI agents capable of making decisions about you, human resources and certain functions of the customer service. You can associate more tools, as an orchestra that assumes the role of responsible for the generation digital tasks based on what you learned by the model.
Jon Siegler, vice president of Platforma and Ia de Servicenowstressed that «The nemotron 15B aaked model, developed by two of the most advanced business companies, has a specific reasoning to boost the next generation of intelligent AI agents. This achieves what generic models cannot, combining business data in real time, the context of the workflow and an advanced reasoning to help AI agents to boost real productivity«.
For its part, Kari Briski, Nvidia generative AI software.he pointed out that together with Servicenow, they have created «An efficient and ready model for use in companies that drives a new class of intelligent agents capable of reasoning to increase equipment productivity. When using the Nvidia data set, NEMOTRON Post-Training and the specific data of the Servicenow domain, Apriel Nemotron 15B, offers advanced reasoning capabilities in a smaller size, which makes it faster, faster and profitable«.
