Terminal-Bench 2.1 (Best Reported Harness)
Terminal-Bench 2.1 (Best Reported Harness) - best score across all evaluation harnesses (Terminus-2, Claude Code, etc.) for Terminal-Bench 2.1.
Category
Agentic Coding
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for Terminal-Bench 2.1 (Best Reported Harness)
| Model | Score | Date | Verified |
|---|---|---|---|
| GLM-5.2 |
95.14%
|
17.06.2026 | Verified |
| Gemini 3.1 Pro |
11.81%
|
17.06.2026 | Unverified |
| GLM-5.1 |
69.00
|
17.06.2026 | Unverified |
| GPT-5.5 |
100.00%
|
17.06.2026 | Unverified |
| Claude Opus 4.8 |
68.75%
|
17.06.2026 | Unverified |