nvidia/Nemotron-Terminal-8B

Nemotron-Terminal-8B is an 8 billion parameter language model developed by NVIDIA, fine-tuned from Qwen3, and specialized for autonomous terminal interaction. It utilizes the Nemotron-Terminal-Corpus dataset to achieve enhanced performance in terminal tasks. This model excels at generating structured JSON outputs for terminal commands and analysis, making it suitable for automating command-line operations. It demonstrates significant accuracy improvements over its Qwen3 baseline on the Terminal-Bench 2.0 benchmark.