Terminal-Bench 3.0

Terminal-Bench 3.0 agentic terminal tasks, evaluated with the Claude Code 2.1.207 harness (reasoning effort=max, 400K context, 128K max output), avg@3 over three rollouts per task scored by the task's official verifier. Open-source SOTA for GLM-5.3 per the GLM-5.3 model card.

Category

Agentic Coding

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for Terminal-Bench 3.0

Model Score Date Verified
Kimi K3
42.67%
28.08.2026 Verified
GLM-5.3
79.00%
28.08.2026 Verified
GLM-5.2
4.60
28.08.2026 Verified
Opus-4.8
55.00%
28.08.2026 Verified
Fable-5
97.00%
28.08.2026 Verified
GPT-5.6 Sol
100.00%
28.08.2026 Verified
DeepSeek-V4.1-Flash
84.67%
10.09.2026 Verified