Structural Integrity

Subcategory of Text Editing. 89 models scored.

Model Leaderboard

All models ranked by their Structural Integrity subcategory score.

# Model Structural Integrity Text Editing Overall
1 Claude Opus 4.6 (Reasoning) 100.00% 98.86% 95.02%
2 Gemini 3.1 Pro (Preview) 100.00% 98.51% 94.37%
3 Claude Sonnet 4.6 (Reasoning) 100.00% 98.30% 93.66%
4 GPT-5 Mini 100.00% 97.13% 92.62%
5 GPT-5.1 100.00% 98.54% 92.54%
6 Claude Opus 4.6 100.00% 98.35% 92.35%
7 GPT-5 100.00% 98.90% 91.93%
8 Qwen 3.5 397B A17B 100.00% 98.05% 91.73%
9 Z.AI GLM 5 100.00% 98.59% 91.23%
10 Claude Sonnet 4.6 100.00% 96.37% 91.15%
11 Qwen 3.5 27B 100.00% 98.69% 90.85%
12 ByteDance Seed 1.6 100.00% 98.40% 90.70%
13 Gemini 3 Flash (Preview, Reasoning) 100.00% 98.12% 90.50%
14 GPT-5.2 100.00% 97.54% 90.26%
15 Claude Opus 4.5 100.00% 97.69% 89.69%
16 Grok 4.1 Fast 100.00% 97.87% 89.55%
17 Z.AI GLM 4.6 100.00% 97.78% 89.11%
18 Gemini 3 Pro (Preview) 100.00% 98.86% 88.79%
19 Claude Sonnet 4 100.00% 99.13% 88.72%
20 Minimax M2.5 100.00% 96.02% 88.71%
21 Z.AI GLM 4.7 100.00% 98.22% 88.69%
22 GPT-4.1 100.00% 94.40% 88.68%
23 Gemini 2.5 Pro 100.00% 98.58% 88.53%
24 Grok 4 100.00% 98.76% 88.12%
25 Claude Sonnet 4.5 100.00% 99.02% 88.03%
26 Claude Opus 4 100.00% 97.25% 87.69%
27 Gemini 2.5 Flash (Reasoning) 100.00% 98.12% 86.51%
28 Grok 4 Fast 100.00% 97.26% 86.15%
29 Qwen 3.5 Plus (2026-02-15) 100.00% 98.10% 85.96%
30 Gemini 2.5 Flash Lite (Reasoning) 100.00% 94.54% 85.75%
31 Mistral Large 3 100.00% 94.09% 85.43%
32 GPT-4o, May 13th (temp=0) 100.00% 95.35% 85.36%
33 Gemini 3 Flash (Preview) 100.00% 97.54% 85.35%
34 Claude Haiku 4.5 100.00% 96.81% 85.14%
35 Claude 3.5 Sonnet 100.00% 96.57% 84.24%
36 GPT-4o, May 13th (temp=1) 100.00% 92.41% 83.80%
37 Claude 3.7 Sonnet 100.00% 97.12% 83.39%
38 GPT-4.1 Mini 100.00% 95.62% 83.20%
39 Hermes 3 405B 100.00% 89.14% 82.86%
40 Mistral Large 2 100.00% 94.16% 82.41%
41 DeepSeek V3.2 100.00% 95.78% 82.25%
42 Gemini 2.5 Flash Lite 100.00% 92.13% 81.08%
43 Gemini 2.5 Flash 100.00% 97.83% 80.60%
44 Mistral Large 100.00% 95.14% 80.15%
45 Writer: Palmyra X5 100.00% 91.20% 79.57%
46 GPT-4o Mini (temp=1) 100.00% 85.78% 79.08%
47 Mistral Small 3.2 24B 100.00% 89.48% 78.60%
48 Gemma 3 12B 100.00% 85.18% 78.41%
49 GPT-4o Mini (temp=0) 100.00% 84.62% 78.29%
50 Gemma 3 27B 100.00% 86.63% 77.85%
51 Mistral Medium 3.1 100.00% 93.77% 77.83%
52 Qwen 2.5 72B 100.00% 89.18% 75.46%
53 Arcee AI: Trinity Large (Preview) 100.00% 86.62% 73.33%
54 Mistral Small Creative 100.00% 90.31% 73.27%
55 Ministral 3 14B 100.00% 86.20% 72.54%
56 GPT-4.1 Nano 100.00% 76.06% 71.94%
57 Ministral 3 8B 100.00% 78.52% 71.76%
58 Gemma 3 4B 100.00% 78.38% 68.57%
59 Ministral 8B 100.00% 77.52% 64.87%
60 Qwen 3.5 122B 98.81% 96.31% 91.53%
61 MoonshotAI: Kimi K2.5 98.81% 97.79% 91.04%
62 o4 Mini High 98.81% 94.36% 90.29%
63 Qwen 3.5 35B 98.81% 94.95% 88.00%
64 Z.AI GLM 4.5 98.81% 95.32% 86.27%
65 DeepSeek V3 (2025-03-24) 98.81% 89.57% 81.99%
66 DeepSeek V3 (2024-12-26) 98.81% 93.58% 83.68%
67 o4 Mini 97.62% 90.61% 88.35%
68 ByteDance Seed 1.6 Flash 97.62% 91.64% 73.27%
69 DeepSeek-V2 Chat 97.62% 90.90% 84.83%
70 Llama 3.1 70B 97.62% 92.10% 78.40%
71 Ministral 3B 97.62% 70.91% 61.29%
72 Qwen 3.5 Flash 96.43% 92.80% 86.38%
73 Z.AI GLM 4.7 Flash 96.43% 85.82% 84.82%
74 GPT-4o, Aug. 6th (temp=0) 96.43% 93.77% 82.45%
75 Ministral 3 3B 96.43% 69.80% 67.22%
76 Cohere Command R+ (Aug. 2024) 95.39% 68.40% 69.03%
77 GPT-5 Nano 95.24% 82.74% 82.60%
78 DeepSeek V3.1 95.24% 87.27% 82.39%
79 Aion 2.0 95.24% 95.34% 89.21%
80 Llama 3.1 Nemotron 70B 95.04% 87.26% 74.70%
81 Llama 3.1 8B 94.05% 75.45% 63.37%
82 Mistral NeMO 93.21% 73.69% 65.04%
83 WizardLM 2 8x22b 93.12% 88.13% 71.07%
84 GPT-4o, Aug. 6th (temp=1) 92.86% 86.72% 82.62%
85 LFM2 24B 90.94% 71.56% 58.77%
86 Arcee AI: Trinity Mini 85.71% 73.88% 70.90%
87 Claude 3 Haiku 81.03% 64.36% 71.19%
88 Hermes 3 70B 78.57% 63.34% 72.57%
89 Rocinante 12B 72.78% 56.31% 54.55%