Terminal-Bench Hard S
AI benchmark · Coding
- Built by
- Laude Institute subset of Terminal-Bench by Stanford University Artificial Analysis
- Domains
- Coding
- Tier
- S why
- Snapshot
What is Terminal-Bench Hard?
Terminal-Bench Hard measures AI agents' ability to complete challenging terminal tasks such as software engineering, system administration, data processing and game playing, verified programmatically.
Coding mostly
“An agentic benchmark evaluating AI capabilities in terminal environments through software engineering, system administration, and data processing tasks.”
artificialanalysis.ai/evaluations/terminalbench-hard ↗“Tasks cover a wide range of engineering, game playing, and system administration tasks that are unlikely to be pattern-matched on training data.”
artificialanalysis.ai/evaluations/terminalbench-hard ↗
Who built Terminal-Bench Hard
Led by Artificial Analysis (subset of Terminal-Bench by Stanford University and the Laude Institute).
Paper: arxiv.org/abs/2601.11868 ↗
Tier S
The standard: frontier labs report it, or a trusted evaluator scores hundreds of models on it.
How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.
- 327 models scored
- in curated RL directories (2 pts)
On X · unclear, low buzz, from 13 posts about Terminal-Bench Hard in the last month
Posts come from one account repeating a claim about Qwen3.8-27B's Terminal-Bench 2.1 gains, with no discussion of the Artificial Analysis Hard slice itself.
How we got to tier S
- Adoption: 327 models scored. Tier S.
- On X: unclear, low buzz. “Posts come from one account repeating a claim about Qwen3.8-27B's Terminal-Bench 2.1 gains, with no discussion of the Artificial Analysis Hard slice itself.” No move.
- In curated RL directories: 2 points, a floor of C.
- Confirmed: a line from its own site or paper places it.
A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.
Where Terminal-Bench Hard sits
Its domain on the map, live. Hover a logo for its name; click to open it. Open it on the full map
Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-11-05 snapshot.