allenai/qwen35-9b-terminaltraj

VISIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 19, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The allenai/qwen35-9b-terminaltraj model is a 9 billion parameter Qwen 3.5-based language model developed by Ai2, fine-tuned using DPPO specifically as a terminal agent. It is optimized for command-line interface tasks, trained on the Terminal-Traj dataset, and features a 32768-token context length. This model is designed to execute and interact within terminal environments, making it suitable for automated scripting and system management applications.

Loading preview...

Overview

allenai/qwen35-9b-terminaltraj is a 9 billion parameter model developed by Ai2, based on the Qwen 3.5 architecture. It has been specifically fine-tuned using DPPO (Distributed Proximal Policy Optimization) to function as a terminal agent. This model is part of a larger collection of terminal agents and was trained as an ablation on the Terminal-Traj dataset.

Key Capabilities

  • Terminal Interaction: Designed for executing commands and interacting within a command-line interface (CLI) environment.
  • DPPO Fine-tuning: Utilizes DPPO for enhanced performance in agent-based tasks.
  • Context Length: Supports a substantial context window of 32768 tokens, allowing for complex command sequences and longer interactions.
  • Evaluation: Achieved 45.8 ± 2.7 on TB Lite and 18.0 ± 0.0 on TB 2.1 benchmarks, demonstrating its proficiency in terminal-based problem-solving.

Use Cases

  • Automated Scripting: Ideal for generating and executing scripts in a terminal.
  • System Management: Can be applied to tasks requiring interaction with operating system commands.
  • Research in Agentic AI: Useful for exploring and developing AI agents that operate in CLI environments.

For more detailed information on the training methodology and evaluation, refer to the associated Tmax paper.