Agentic terminal coding
Terminal-Bench 2.1
Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better.
Rankings
Higher is better188.8%
GPT-5.6 Sol
OpenAI · Jun 26 2026
288.3%
Kimi K3
Moonshot AI · Jul 16 2026
388%
Claude Fable 5
Anthropic · Jun 9 2026
484.3%
GPT-5.6 Terra
OpenAI · Jun 26 2026
583.3%
Grok 4.5
SpaceXAI · Jul 8 2026
682.5%
GPT-5.6 Luna
OpenAI · Jun 26 2026
781%
GLM-5.2
Z.ai · Jun 16 2026
880.4%
Claude Sonnet 5
Anthropic · Jun 30 2026
980%
Muse Spark 1.1
Meta · Jul 9 2026
1078.2%
GPT-5.5
OpenAI · Apr 23 2026
1176.2%
Gemini 3.5 Flash
Google · May 19 2026
1274.6%
Claude Opus 4.8
Anthropic · May 28 2026
1373%
Composer 2.5
Cursor · May 18 2026
1470.3%
Gemini 3.1 Pro
Google · Feb 19 2026
1567.3%
Muse Spark
Meta · Apr 8 2026
1666.1%
Claude Opus 4.7
Anthropic · Apr 16 2026
1758%
Gemini 3.0 Flash
Google · Dec 17 2025