benchgap
General

SEC-Bench Pro leaderboard

As of 2026-11-21, the highest measured score on SEC-Bench Pro is 88.3% by GPT-6.2 Sol. 100 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 5.295.1%estimated ± 6.9 pp, low confidence
2GPT-6.2 Sol88.3%measured
3Gemini 4 Carbon88.2%measured
4GPT-6 Cyber87.0%measured
5GPT-6.2 Astra86.8%measured
6Gemini 4 Neon86.1%estimated ± 6.9 pp, medium confidence
7GPT-6 Astra85.4%measured
8Gemini 3.8 Flash85.2%estimated ± 5.6 pp, medium confidence
9Claude Opus 5.585.0%estimated ± 5.6 pp, medium confidence
10Gemini 3.6 Flash84.1%estimated ± 5.6 pp, medium confidence
11Gemini 4 Argon83.4%measured
12GPT-6.1 Terra83.3%measured
13Claude Mythos 5.283.3%estimated ± 3.9 pp, high confidence
14Qwen3.8 Max82.8%estimated ± 5.6 pp, medium confidence
15Claude Sonnet 5.581.7%estimated ± 5.6 pp, medium confidence
16Gemini 3.1 Pro81.7%estimated ± 5.6 pp, medium confidence
17Gemini 3 Flash80.8%estimated ± 5.6 pp, medium confidence
18Claude Opus 580.2%estimated ± 5.8 pp, medium confidence
19GPT-6.1 Sol78.8%measured
20Claude Mythos 578.4%estimated ± 3.9 pp, high confidence
21Kimi K3.1 Pro77.3%estimated ± 3.9 pp, high confidence
22Claude Fable 5.175.5%estimated ± 6.9 pp, medium confidence
23Claude Mythos Preview75.2%estimated ± 3.9 pp, high confidence
24GPT-5.5 Pro71.6%estimated ± 12.3 pp, low confidence
25GPT-5.6 Sol71.2%measured
26GPT-5.4 Pro69.3%estimated ± 12.3 pp, low confidence
27GLM-5.368.6%estimated ± 3.9 pp, high confidence
28Inkling 2 Small68.1%estimated ± 6.9 pp, medium confidence
29GPT-5.6 Terra67.8%estimated ± 3.9 pp, high confidence
30GPT-6 Sol66.3%measured
31MiMo-V2.6-Pro66.3%measured
32Mistral Large 465.4%estimated ± 6.9 pp, low confidence
33DeepSeek V4.1 Flash 101762.8%measured
34Gemini 4 Hydrogen61.2%estimated ± 6.9 pp, low confidence
35Qwen3.8-27B61.2%estimated ± 9.3 pp, medium confidence
36Claude Fable 560.3%estimated ± 7.5 pp, medium confidence
37Gemma 4 31B56.5%estimated ± 5.6 pp, medium confidence
38SWE-2 Flash55.7%estimated ± 3.9 pp, high confidence
39GPT-5.6 Luna54.3%estimated ± 3.9 pp, high confidence
40Gemini 3.7 Flash53.9%estimated ± 9.3 pp, medium confidence
41MiniMax M3.1 Flash53.4%estimated ± 9.3 pp, medium confidence
42Kimi K353.2%estimated ± 3.9 pp, high confidence
43Muse Wonder52.3%estimated ± 7.5 pp, low confidence
44DeepSeek V4 Flash 073151.8%estimated ± 9.3 pp, medium confidence
45Muse Spark Open51.8%estimated ± 9.3 pp, medium confidence
46Hy4 preview51.6%estimated ± 9.3 pp, medium confidence
47DeepSeek V4 Pro50.2%estimated ± 9.3 pp, medium confidence
48Claude Opus 4.850.1%estimated ± 5.8 pp, medium confidence
49Ling 3.1 Flash49.7%estimated ± 9.3 pp, medium confidence
50o348.5%estimated ± 9.3 pp, medium confidence
51Grok 4.748.2%estimated ± 5.6 pp, medium confidence
52MiMo-V2.6-Flash47.5%measured
53Muse Spark 1.146.8%estimated ± 12.3 pp, low confidence
54Claude Sonnet 546.4%estimated ± 5.8 pp, medium confidence
55Muse Spark 1.346.0%estimated ± 9.3 pp, medium confidence
56GPT-5.545.8%measured
57GPT-5.445.4%estimated ± 7.5 pp, low confidence
58Step 5 Preview45.4%estimated ± 9.3 pp, medium confidence
59MiniMax M344.9%estimated ± 9.3 pp, medium confidence
60Qwen3.5 Flash42.5%estimated ± 9.3 pp, medium confidence
61GPT-5.242.4%estimated ± 7.5 pp, low confidence
62Qwen 3.6 Max (preview)41.8%estimated ± 9.3 pp, medium confidence
63Grok 4.2041.3%estimated ± 12.3 pp, low confidence
64Qwen3.5 397B41.1%estimated ± 9.3 pp, medium confidence
65Qwen3.5-122B-A10B40.4%estimated ± 9.3 pp, medium confidence
66Qwen3.5 Plus40.4%estimated ± 9.3 pp, medium confidence
67Qwen3.7 Plus40.4%estimated ± 9.3 pp, medium confidence
68Claude Sonnet 4.639.8%estimated ± 9.3 pp, medium confidence
69GPT-5.139.8%estimated ± 9.3 pp, medium confidence
70Gemma 4 26B A4B39.5%estimated ± 5.6 pp, medium confidence
71Qwen3.7 Flash39.1%estimated ± 9.3 pp, medium confidence
72Muse Spark 1.239.1%estimated ± 12.3 pp, low confidence
73Grok 4.638.3%estimated ± 5.6 pp, medium confidence
74Seed 2.1 Turbo37.9%estimated ± 12.3 pp, low confidence
75GPT-4o mini37.1%estimated ± 9.3 pp, medium confidence
76Qwen3.6 Plus37.1%estimated ± 9.3 pp, medium confidence
77Qwen3.6-27B37.0%estimated ± 9.3 pp, low confidence
78Claude Opus 4.736.9%estimated ± 7.5 pp, low confidence
79Beam 1.0 501B-A23B36.4%estimated ± 9.3 pp, low confidence
80Claude Haiku 5.535.8%estimated ± 5.6 pp, medium confidence
81GLM-5.3-Flash34.4%estimated ± 9.3 pp, medium confidence
82GPT-6 Luna34.2%measured
83GPT-4.1 mini33.7%estimated ± 9.3 pp, medium confidence
84GPT-5.4 mini33.7%estimated ± 9.3 pp, medium confidence
85o3-mini33.7%estimated ± 9.3 pp, medium confidence
86Grok 4.533.0%estimated ± 10.7 pp, low confidence
87GPT-5.4 nano32.4%estimated ± 9.3 pp, low confidence
88o4-mini32.4%estimated ± 9.3 pp, low confidence
89Qwen3 Max32.4%estimated ± 9.3 pp, low confidence
90GLM-4.731.8%estimated ± 12.3 pp, low confidence
91Inkling31.8%estimated ± 12.3 pp, low confidence
92Seed 2.1 Pro31.8%estimated ± 12.3 pp, low confidence
93Inkling-Small30.8%estimated ± 9.3 pp, low confidence
94GLM-5.130.5%estimated ± 12.3 pp, low confidence
95Claude Opus 4.529.6%estimated ± 7.5 pp, low confidence
96GPT-OSS 120B29.0%estimated ± 9.3 pp, low confidence
97Grok 4.327.8%estimated ± 12.3 pp, low confidence
98Claude Opus 4.626.9%estimated ± 7.5 pp, low confidence
99GPT-526.9%estimated ± 7.5 pp, low confidence
100Kimi K2.7 Code22.4%estimated ± 12.3 pp, low confidence
101Qwen3.6-35B-A3B22.3%estimated ± 9.3 pp, low confidence
102GLM-5.221.1%estimated ± 7.5 pp, low confidence
103Qwen3.7 Max21.1%estimated ± 7.5 pp, low confidence
104Gemini 3.5 Flash-Lite20.7%estimated ± 5.6 pp, low confidence
105GPT-4.118.0%estimated ± 12.3 pp, low confidence
106Nemotron 3 Ultra17.2%estimated ± 9.3 pp, low confidence
107o116.5%estimated ± 12.3 pp, low confidence
108GPT-4o15.0%estimated ± 12.3 pp, low confidence
109MAI-Thinking-113.9%estimated ± 9.3 pp, low confidence
110Claude Haiku 4.513.4%estimated ± 12.3 pp, low confidence
111Gemini 3.5 Flash11.5%estimated ± 7.5 pp, low confidence
112Claude 3 Opus6.9%estimated ± 12.3 pp, low confidence
113Claude Sonnet 4.56.0%estimated ± 7.5 pp, low confidence
114Kimi K2.66.0%estimated ± 7.5 pp, low confidence
115GPT-4.1 nano0.0%estimated ± 12.3 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General