An Artificial Analysis benchmark variant focused on difficult tool-use tasks in terminal environments.
Browse the latest scores, model modes, release dates, and parameter sizes for Terminal Bench Hard.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 Opus 4.5 Extended ThinkingTools | 44.00 | 2025-11-25 | Unknown | Closed | |
 GPT-5.1 Thinking Level · HighTools | 43.00 | 2025-11-12 | Unknown | Closed | |
 GLM-5 Thinking EnabledTools | 43.00 | 2026-02-11 | 744B | Free Commercial | |
4 |  Gemini 3.0 Pro (Preview 11-2025) Thinking Level · HighTools | 42.00 | 2025-11-18 | Unknown | Closed |
5 | Gemini 3.0 Pro (Preview 11-2025) Thinking EnabledTools | 39.00 | 2025-11-18 | Unknown | Closed |
6 |  MiniMax-M2.7 Thinking EnabledTools | 39.00 | 2026-03-18 | 229B | Non-Commercial |
7 | GLM-4.7 Thinking EnabledTools | 33.30 | 2025-12-22 | 358B | Free Commercial |
8 | Claude Sonnet 4.5 Thinking EnabledTools | 33.00 | 2025-09-30 | Unknown | Closed |
9 | Opus 4.1 Extended ThinkingTools | 32.00 | 2025-08-06 | Unknown | Closed |
10 | GLM-4.7-Flash Thinking EnabledTools | 32.00 | 2026-01-19 | 31B | Free Commercial |
11 | GLM-4.7-Flash Standard ModeTools | 30.00 | 2026-01-19 | 31B | Free Commercial |
12 | Gemini 2.5-Pro Thinking EnabledTools | 25.00 | 2025-06-05 | Unknown | Closed |
13 | Claude Sonnet 3.7 Thinking EnabledTools | 21.00 | 2025-02-25 | Unknown | Closed |
