benchgap
Agentic · tools

AA EnterpriseOps-Gym leaderboard

As of 2026-11-22, the highest measured score on AA EnterpriseOps-Gym is 51.1% by Claude Fable 5. 57 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 551.1%measured
2Kimi K3.1 Pro50.6%estimated ± 7.2 pp, low confidence
3Gemini 3.5 Flash50.1%measured
4DeepSeek V4 Pro49.6%measured
5Qwen4 Max48.5%measured
6Grok 4.648.3%measured
7Mistral Large 447.9%estimated ± 7.2 pp, low confidence
8GPT-6 Luna47.6%estimated ± 7.2 pp, low confidence
9Claude Opus 547.5%measured
10Muse Spark 1.247.3%measured
11Step 5 Preview47.2%measured
12Muse Wonder Full47.2%estimated ± 1.2 pp, low confidence
13Muse Wonder47.0%estimated ± 1.2 pp, medium confidence
14GLM-5.3-Flash47.0%estimated ± 7.2 pp, low confidence
15Grok 4.747.0%estimated ± 7.2 pp, low confidence
16LongCat-2.546.8%estimated ± 1.2 pp, medium confidence
17Muse Spark 1.346.8%estimated ± 1.2 pp, medium confidence
18Inkling 2 Small46.8%measured
19Ultramarine 1 Flash46.8%estimated ± 1.2 pp, medium confidence
20MiMo-V2.6-Pro46.7%estimated ± 1.2 pp, medium confidence
21Muse Spark Open46.7%estimated ± 1.2 pp, medium confidence
22Hy4 preview46.7%estimated ± 1.2 pp, medium confidence
23MiMo-V2.6-Flash46.7%estimated ± 1.2 pp, medium confidence
24GPT-5.546.6%measured
25Qwen3.8-Flash-Next46.4%estimated ± 1.2 pp, medium confidence
26Muse Spark 1.146.4%estimated ± 1.2 pp, medium confidence
27Qwen3.8 Max46.3%estimated ± 1.2 pp, medium confidence
28Atria Dawn Preview46.1%estimated ± 1.2 pp, medium confidence
29Inkling 2 Mini46.1%estimated ± 1.2 pp, medium confidence
30Claude Opus 4.745.8%estimated ± 1.2 pp, medium confidence
31Step 3.7 Flash45.8%estimated ± 2.3 pp, medium confidence
32Kimi K2.645.8%estimated ± 2.3 pp, medium confidence
33Apodex 1.145.8%estimated ± 2.3 pp, medium confidence
34dots3-note Preview45.7%estimated ± 2.3 pp, medium confidence
35Ling 3.0 Flash Sante45.5%estimated ± 2.3 pp, medium confidence
36Kimi K345.3%measured
37GPT-5.445.2%estimated ± 1.2 pp, medium confidence
38Claude Sonnet 4.645.0%estimated ± 1.2 pp, medium confidence
39Claude Opus 4.645.0%estimated ± 1.2 pp, medium confidence
40Qwen3.7 Max45.0%measured
41GPT-5.244.7%estimated ± 1.2 pp, medium confidence
42Claude Sonnet 544.7%measured
43GPT-5.3 Codex44.7%estimated ± 1.2 pp, medium confidence
44Claude Opus 4.544.5%estimated ± 1.2 pp, low confidence
45Qwen3.8-27B44.2%measured
46Beam 1.0 501B-A23B NVFP444.2%estimated ± 2.3 pp, medium confidence
47Beam 1.0 501B-A23B44.1%estimated ± 2.3 pp, medium confidence
48Claude Opus 4.844.0%measured
49Beam 1.0 501B-A23B FP844.0%estimated ± 2.3 pp, medium confidence
50Claude Sonnet 4.543.7%estimated ± 1.2 pp, low confidence
51GPT-5.1-Codex43.5%estimated ± 1.2 pp, low confidence
52GPT-5.2-Codex43.4%estimated ± 1.2 pp, low confidence
53Muse Spark43.2%estimated ± 2.3 pp, medium confidence
54Muse Glimmer 30B43.1%estimated ± 2.3 pp, medium confidence
55GPT-5.6 Sol42.9%measured
56GLM-5.242.7%measured
57Inkling-Small42.7%measured
58Mercury 342.6%estimated ± 2.3 pp, medium confidence
59Claude 4.1 Opus42.5%estimated ± 1.2 pp, low confidence
60Gemini 3.5 Flash-Lite42.3%measured
61Gemini 3.1 Pro42.2%measured
62Qwen3.5 Plus41.4%estimated ± 1.2 pp, low confidence
63Claude 4 Sonnet41.3%estimated ± 1.2 pp, low confidence
64Grok 4.2040.9%estimated ± 2.3 pp, low confidence
65Claude Haiku 4.540.4%estimated ± 1.2 pp, low confidence
66MiniMax M338.2%estimated ± 7.2 pp, low confidence
67DeepSeek V4.1 Flash 101737.9%estimated ± 7.2 pp, low confidence
68Gemini 3 Flash37.6%estimated ± 1.2 pp, low confidence
69Gemini 3 Pro37.6%estimated ± 1.2 pp, low confidence
70GLM-5.336.4%measured
71Nemotron 3 Ultra35.8%estimated ± 7.2 pp, low confidence
72Gemini 3.8 Flash35.4%estimated ± 7.2 pp, low confidence
73Kimi K2.535.0%estimated ± 1.2 pp, low confidence
74GPT-534.8%estimated ± 1.2 pp, low confidence
75Claude Fable 5.134.0%estimated ± 7.2 pp, low confidence
76Claude Opus 5.534.0%estimated ± 7.2 pp, low confidence
77GPT-6 Astra33.9%estimated ± 7.2 pp, low confidence
78GPT-6 Sol33.9%estimated ± 7.2 pp, low confidence
79GPT-6.2 Astra29.9%measured
80GPT-6.2 Sol28.9%measured

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General