nvidia/Nemotron-Terminal-32B
TEXT GENERATIONConcurrency Cost:2Model Size:32BQuant:FP8Ctx Length:32kPublished:Feb 17, 2026License:otherArchitecture:Transformer0.0K Warm
Nemotron-Terminal-32B is a 32 billion parameter model developed by NVIDIA, fine-tuned from Qwen3, and specialized for autonomous terminal interaction. It leverages the Nemotron-Terminal-Corpus dataset to achieve enhanced performance in terminal tasks, significantly outperforming its Qwen3 baseline. This model is designed for use cases requiring automated command-line operations and structured JSON output for terminal control.
Loading preview...
Nemotron-Terminal-32B: Specialized for Autonomous Terminal Interaction
Nemotron-Terminal-32B is a 32 billion parameter model from NVIDIA, part of the Nemotron-Terminal family, which is specifically fine-tuned from the Qwen3 architecture. This model is engineered for autonomous terminal interaction, utilizing the extensive Nemotron-Terminal-Corpus dataset for training.
Key Capabilities & Performance
- Specialized Terminal Control: Designed to understand and execute tasks within a terminal environment.
- Enhanced Accuracy: Demonstrates a significant leap in capability on the Terminal-Bench 2.0 benchmark, achieving 27.4% accuracy compared to its Qwen3 baseline's 3.37%.
- Structured Output: Outputs responses in a structured JSON format, including analysis, plan, and a list of commands with keystrokes and duration, facilitating programmatic interaction.
- Terminus 2 Scaffolding: Trained and optimized for use with the Terminus 2 scaffolding, ensuring consistent performance.
Use Cases
- Automated System Administration: Ideal for automating complex command-line tasks and system management.
- Developer Tools: Can be integrated into development workflows for automated testing, deployment, or environment setup.
- Intelligent Agents: Suitable for building agents that can interact with and control terminal-based applications autonomously.