benchgap
Agentic · terminal

Terminal-Bench 2.0 leaderboard

As of 2026-11-18, the highest measured score on Terminal-Bench 2.0 is 82.0% by GPT-5.5. 53 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1GPT-5.6 Sol86.0%estimated ± 5.4 pp, low confidence
2Claude Opus 5.583.3%estimated ± 5.8 pp, low confidence
3GPT-6 Astra83.1%estimated ± 5.8 pp, low confidence
4GPT-5.582.0%measured
5Claude Fable 581.9%estimated ± 5.4 pp, low confidence
6Claude Fable 5.181.9%estimated ± 5.8 pp, low confidence
7Claude Opus 581.6%estimated ± 5.8 pp, low confidence
8Claude Sonnet 5.580.8%estimated ± 5.8 pp, low confidence
9GPT-6 Sol80.8%estimated ± 5.8 pp, low confidence
10Gemini 3.8 Flash79.8%estimated ± 5.8 pp, low confidence
11Kimi K379.6%estimated ± 5.8 pp, low confidence
12GPT-5.6 Luna78.5%estimated ± 5.8 pp, low confidence
13Grok 4.678.1%estimated ± 5.8 pp, low confidence
14Gemini 3.7 Flash77.6%estimated ± 5.8 pp, low confidence
15GPT-5.6 Terra77.6%estimated ± 5.8 pp, low confidence
16GPT-5.3 Codex77.3%measured
17MiMo-V2.6-Flash77.0%estimated ± 5.8 pp, medium confidence
18Claude Opus 4.875.9%estimated ± 5.4 pp, medium confidence
19Claude Sonnet 575.8%estimated ± 5.8 pp, medium confidence
20DeepSeek V4.1 Flash 101775.8%estimated ± 5.8 pp, medium confidence
21Gemini 3.5 Flash75.6%estimated ± 5.8 pp, medium confidence
22Gemini 3.6 Flash75.4%estimated ± 5.8 pp, medium confidence
23Grok 4.775.1%estimated ± 5.8 pp, medium confidence
24GPT-5.475.1%measured
25GPT-6 Luna74.9%estimated ± 5.8 pp, medium confidence
26Muse Spark 1.374.5%estimated ± 5.8 pp, medium confidence
27GLM-5.374.0%estimated ± 5.8 pp, medium confidence
28Muse Spark 1.272.8%estimated ± 5.8 pp, medium confidence
29Muse Spark 1.172.6%estimated ± 5.8 pp, medium confidence
30Grok 4.571.6%estimated ± 5.8 pp, medium confidence
31MiMo-V2.6-Pro71.6%estimated ± 5.8 pp, medium confidence
32Qwen3.8 Max71.3%estimated ± 5.8 pp, medium confidence
33Gemini 3.1 Pro70.7%estimated ± 5.4 pp, medium confidence
34Qwen3.7 Plus70.3%measured
35Qwen3.7 Max69.7%measured
36Claude Opus 4.769.4%measured
37Composer 2.569.3%measured
38MiMo-V2.5-Pro68.4%measured
39GLM-5.3-Flash68.3%estimated ± 5.8 pp, medium confidence
40DeepSeek V4 Pro67.9%measured
41GLM-5.267.8%estimated ± 5.4 pp, medium confidence
42Kimi K2.666.7%measured
43MiMo-V2.565.8%measured
44Claude Opus 4.665.4%measured
45Qwen 3.6 Max (preview)65.4%measured
46Qwen3.8-27B65.1%estimated ± 5.8 pp, medium confidence
47Claude Opus 4.5 Thinking64.8%estimated ± 5.4 pp, medium confidence
48GPT-5.264.8%estimated ± 5.4 pp, medium confidence
49Gemini 4 Argon64.6%estimated ± 5.8 pp, medium confidence
50GPT-5.163.9%estimated ± 5.4 pp, medium confidence
51GLM-5.163.5%measured
52Kimi K2.7 Code63.4%estimated ± 5.4 pp, medium confidence
53Hy4 preview62.6%estimated ± 5.8 pp, medium confidence
54Inkling-Small62.6%estimated ± 5.8 pp, medium confidence
55Gemini 3 Flash61.7%estimated ± 5.8 pp, medium confidence
56Composer 261.7%measured
57Qwen3.6 Plus61.6%measured
58MiniMax M361.5%estimated ± 5.8 pp, medium confidence
59GPT-5.4 mini60.0%measured
60Nemotron 3 Ultra59.4%estimated ± 5.8 pp, medium confidence
61Solar Open 259.3%estimated ± 5.4 pp, low confidence
62Claude Opus 4.559.3%measured
63Qwen3.6-27B59.3%measured
64Claude Sonnet 4.659.1%measured
65Muse Spark59.0%measured
66Gemini 3.5 Flash-Lite58.8%estimated ± 5.8 pp, medium confidence
67Ling 3.0 Flash58.8%estimated ± 5.8 pp, medium confidence
68MiniMax M2.757.0%measured
69DeepSeek V4 Flash 073156.9%measured
70Inkling56.7%estimated ± 5.8 pp, medium confidence
71GLM-556.2%measured
72Hy3 Preview54.4%measured
73Claude Haiku 4.553.4%estimated ± 5.8 pp, medium confidence
74Qwen3.5 397B52.5%measured
75Grok 4.351.8%estimated ± 5.8 pp, medium confidence
76Qwen3.6-35B-A3B51.5%measured
77Kimi K2.550.8%measured
78Claude Sonnet 4.550.0%measured
79Qwen3.5-122B-A10B49.4%measured
80Mistral Medium 3.5 128B49.1%estimated ± 5.8 pp, medium confidence
81Grok 4.2047.1%measured
82GPT-5.4 nano46.3%measured
83MAI-Thinking-146.0%measured
84Laguna M.145.8%measured
85Mercury 2.544.8%estimated ± 5.8 pp, medium confidence
86Gemini 3.1 Flash-Lite44.4%estimated ± 5.8 pp, medium confidence
87Qwen3.5-27B41.6%measured
88GLM-4.741.0%measured
89Qwen3.5-35B-A3B40.5%measured
90Laguna XS 2.137.5%measured
91Laguna XS.235.7%measured
92LongCat-Flash-Lite-Sparse33.7%measured
93Command A+25.9%estimated ± 5.8 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General