Our goal with the Neurometric leaderboard is to show how much models vary on different tasks. Since launch we are working to build the best leaderboard for “systems 2” AI. As part of that process of adding more models, more tasks, and more thinking algorithms, we are excited to announce today that we support NVIDIA Nemotron models. Below is a sample chart showing NVIDIA Nemotron 3.30b on one of the CRM-Arena tasks.
You can see the Nemotron model wins on the “lead qualification” task. If this is a task you are building into an agent, you may want to consider the Nemotron model.
If you want to run these tests on your own data to see what models work best, you can do so by inserting one line of code into your application. Check out our homepage to copy and paste that code snippet.