benchgap
Agentic · terminal

Terminal-Bench 4.1 leaderboard

As of 2026-11-18, the highest measured score on Terminal-Bench 4.1 is 65.8% by Claude Fable 5.2. 103 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 5.265.8%measured
2Claude Opus 5.565.8%measured
3GPT-6.2 Astra65.0%estimated ± 5.5 pp, low confidence
4Gemini 4 Carbon64.6%measured
5Claude Mythos 5.264.6%estimated ± 5.0 pp, medium confidence
6Claude Sonnet 5.563.9%measured
7GPT-6.2 Sol63.2%estimated ± 5.5 pp, medium confidence
8GPT-6.1 Sol60.9%measured
9Claude Mythos 5.160.9%estimated ± 5.0 pp, medium confidence
10GPT-6 Astra60.3%measured
11Kimi K3.1 Pro59.7%measured
12Claude Fable 5.157.6%measured
13Muse Wonder Full56.5%estimated ± 6.4 pp, medium confidence
14Gemini 4 Hydrogen56.4%measured
15Gemini 4 Argon55.5%measured
16Claude Opus 553.3%measured
17GPT-6.1 Terra53.3%measured
18Pareto 26.951.7%estimated ± 5.0 pp, medium confidence
19Pareto 26.10 Preview51.4%estimated ± 5.0 pp, medium confidence
20GPT-6 Sol50.3%measured
21Qwen4 Max48.3%estimated ± 5.0 pp, medium confidence
22Muse Wonder47.0%measured
23ERNIE 5.5 Flash Preview45.4%estimated ± 5.0 pp, medium confidence
24Claude Fable 544.9%measured
25Ling 3.1 Flash44.7%estimated ± 5.0 pp, medium confidence
26GPT-5.541.4%estimated ± 9.2 pp, low confidence
27GLM-5.341.2%measured
28Claude Haiku 5.540.0%measured
29Grok 4.738.8%measured
30DeepSeek V4.1 Flash 101738.5%measured
31GPT-5.6 Sol38.5%measured
32Hy437.9%measured
33GLM-5.3-Flash37.0%measured
34LongCat-2.532.9%estimated ± 5.0 pp, medium confidence
35MiniMax M3.1 Flash32.3%estimated ± 5.0 pp, medium confidence
36Muse Ember30.6%measured
37Mistral Large 427.6%measured
38MiMo-V2.6-Pro27.0%measured
39Step 5 Preview26.0%estimated ± 5.0 pp, medium confidence
40Claude Mythos 524.9%estimated ± 10.9 pp, low confidence
41Claude Opus 4.823.6%measured
42Qwen3.8 Max23.5%estimated ± 10.9 pp, low confidence
43Kimi K323.0%measured
44Gemini 3.6 Flash22.8%estimated ± 14.9 pp, low confidence
45Hy4 preview22.3%estimated ± 10.9 pp, low confidence
46Ember-121.8%measured
47MAI Code 221.6%estimated ± 10.9 pp, low confidence
48GPT-5.6 Terra21.5%measured
49SWE-220.3%estimated ± 5.0 pp, medium confidence
50Ultramarine 1 Flash20.2%estimated ± 10.9 pp, low confidence
51Muse Spark 1.219.9%estimated ± 10.9 pp, low confidence
52DeepSeek V4 Flash 073119.7%estimated ± 10.9 pp, low confidence
53Grok 4.619.7%measured
54Sakana Fugu-Ultra19.1%estimated ± 10.9 pp, low confidence
55SWE-1.718.6%estimated ± 10.9 pp, low confidence
56Gemini 3.8 Flash18.2%measured
57GPT-5.6 Luna17.6%measured
58Ripple-65B Preview17.6%estimated ± 10.9 pp, low confidence
59Claude Opus 4.717.4%estimated ± 14.9 pp, low confidence
60Sakana Fugu17.3%estimated ± 10.9 pp, low confidence
61Muse Spark 1.117.1%estimated ± 10.9 pp, low confidence
62Particle-117.0%estimated ± 10.9 pp, low confidence
63Inkling 2 Small16.4%estimated ± 6.4 pp, medium confidence
64SWE-2 Flash16.3%estimated ± 5.0 pp, medium confidence
65MiMo-V2.6-Flash16.1%measured
66Beam 1.0 501B-A23B FP816.0%estimated ± 10.9 pp, low confidence
67Hy4 flash16.0%estimated ± 5.0 pp, medium confidence
68Beam 1.0 501B-A23B NVFP415.6%estimated ± 10.9 pp, low confidence
69Atria Dawn Preview15.5%estimated ± 10.9 pp, low confidence
70MAI Code 2 Flash15.5%estimated ± 10.9 pp, low confidence
71GPT-6 Luna15.2%measured
72Muse Spark 1.314.9%measured
73Ornith-1.0-397B14.7%estimated ± 10.9 pp, low confidence
74Gemini 3.5 Flash13.4%estimated ± 10.9 pp, low confidence
75Gemini 3.7 Flash12.4%measured
76Grok 4.512.4%measured
77dots3-note Preview12.4%estimated ± 10.9 pp, low confidence
78MiMo-V2.512.2%estimated ± 14.9 pp, low confidence
79Claude Sonnet 512.1%measured
80GPT-5.411.1%estimated ± 9.2 pp, low confidence
81Jais-3-375B-A23B11.1%estimated ± 10.9 pp, low confidence
82Composer 2.511.1%estimated ± 14.9 pp, low confidence
83Ornith-1.5-397B11.1%estimated ± 4.8 pp, medium confidence
84Claude Sonnet 4.610.5%estimated ± 14.9 pp, low confidence
85GLM-5.110.4%estimated ± 14.9 pp, low confidence
86OrcaCyber-Nano-1.09.8%estimated ± 10.9 pp, low confidence
87Qwen3.8-27B9.6%estimated ± 6.4 pp, medium confidence
88Gemini 3 Flash9.2%estimated ± 14.9 pp, low confidence
89Seed 2.1 Pro8.4%estimated ± 10.9 pp, low confidence
90Apodex 1.18.2%estimated ± 10.9 pp, low confidence
91Laguna S 2.17.6%estimated ± 10.9 pp, low confidence
92MiniMax M37.6%measured
93MiniMax M2.77.4%estimated ± 14.9 pp, low confidence
94Quasar 438B6.8%estimated ± 10.9 pp, low confidence
95DeepSeek V4 Pro6.7%measured
96Beam 1.0 501B-A23B6.4%measured
97Grok 4.206.2%estimated ± 14.9 pp, low confidence
98Claude Haiku 4.56.1%estimated ± 14.9 pp, low confidence
99Grok 4.35.6%estimated ± 14.9 pp, low confidence
100GPT-5.4 nano5.5%estimated ± 14.9 pp, low confidence
101Seed 2.1 Turbo5.1%estimated ± 10.9 pp, low confidence
102Mistral Medium 3.5 128B5.0%estimated ± 14.9 pp, low confidence
103Jais-3-MoVA-36B-A4B4.8%estimated ± 10.9 pp, low confidence
104Mercury 2.54.1%estimated ± 14.9 pp, low confidence
105Gemini 3.1 Flash-Lite4.0%estimated ± 14.9 pp, low confidence
106Laguna M.14.0%estimated ± 14.9 pp, low confidence
107Gemini 3.1 Pro2.7%measured
108Pokee-Isaac 28B2.7%estimated ± 10.9 pp, low confidence
109Laguna XS.22.7%estimated ± 14.9 pp, low confidence
110Qwen3.7 Max2.2%estimated ± 9.2 pp, low confidence
111Claude Opus 4.62.2%estimated ± 9.2 pp, low confidence
112Claude Opus 4.5 Thinking2.2%estimated ± 9.2 pp, low confidence
113GPT-5.22.2%estimated ± 9.2 pp, low confidence
114Qwen3.7 Plus2.2%estimated ± 9.2 pp, low confidence
115GPT-5.12.2%estimated ± 9.2 pp, low confidence
116Muse Spark2.2%estimated ± 9.2 pp, low confidence
117Kimi K2.7 Code2.2%estimated ± 9.2 pp, low confidence
118Kimi K2.62.2%estimated ± 9.2 pp, low confidence
119Qwen 3.6 Max (preview)2.2%estimated ± 9.2 pp, low confidence
120Qwen3.6 Plus2.2%estimated ± 9.2 pp, low confidence
121GLM-52.2%estimated ± 9.2 pp, low confidence
122MiMo-V2.5-Pro2.2%estimated ± 9.2 pp, low confidence
123Ornith-1.0-35B1.8%estimated ± 10.9 pp, low confidence
124GPT-5.4 mini1.8%measured
125Command A+1.6%estimated ± 14.9 pp, low confidence
126Jais-3-32B1.3%estimated ± 10.9 pp, low confidence
127Inkling-Small0.9%measured
128MAI-Code-1.1-Flash0.6%estimated ± 10.9 pp, low confidence
129Nemotron 3.5 Super VL 120B A12B0.3%measured
130A.X K20.0%estimated ± 10.9 pp, low confidence
131Flint0.0%estimated ± 10.9 pp, low confidence
132Gemini 3.5 Flash-Lite0.0%estimated ± 10.9 pp, low confidence
133GLM-5.20.0%estimated ± 4.8 pp, medium confidence
134Inkling0.0%measured
135Jais-3-3.7B0.0%estimated ± 10.9 pp, low confidence
136Jais-3-7B0.0%estimated ± 10.9 pp, low confidence
137K-EXAONE 2.00.0%estimated ± 10.9 pp, low confidence
138Ling 3.0 Flash0.0%estimated ± 10.9 pp, low confidence
139Muse Glimmer 30B0.0%estimated ± 10.9 pp, low confidence
140Nemotron 3.5 Bolt 74B-A9B0.0%estimated ± 10.9 pp, low confidence
141Nemotron 3.5 Bolt 74B-A9B BF160.0%estimated ± 10.9 pp, low confidence
142Nemotron 3.5 Super VL 120B A12B NVFP40.0%estimated ± 10.9 pp, low confidence
143Nemotron 3 Ultra0.0%estimated ± 10.9 pp, low confidence
144Ornith-1.0-9B0.0%estimated ± 10.9 pp, low confidence
145Ornith-1.5-35B-A3B0.0%estimated ± 4.8 pp, medium confidence
146Ornith-1.5-9B0.0%estimated ± 10.9 pp, low confidence
147Solar Pro 40.0%estimated ± 10.9 pp, low confidence
148Step 3.7 Flash0.0%estimated ± 10.9 pp, low confidence
149Ternary Bonsai 2 27B0.0%estimated ± 10.9 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General