Model Leaderboard

All models ranked by their Dialogue subcategory score.

# Model Dialogue Creative Writing Overall
1 Claude Sonnet 4.6 95.20% 83.31% 91.15%
2 Claude Sonnet 4.6 (Reasoning) 94.92% 83.09% 93.66%
3 Claude Opus 4.6 (Reasoning) 93.56% 84.55% 95.02%
4 Qwen 3.5 397B A17B 92.54% 86.93% 91.73%
5 Claude Opus 4.6 92.37% 83.59% 92.35%
6 GPT-5 91.30% 86.87% 91.93%
7 Claude Sonnet 4.5 89.63% 84.19% 88.03%
8 Qwen 3.5 35B 89.61% 83.51% 88.00%
9 Claude Opus 4 89.08% 83.79% 87.69%
10 Minimax M2.5 88.93% 81.21% 88.71%
11 Claude Opus 4.5 88.91% 81.71% 89.69%
12 Qwen 3.5 27B 88.67% 82.54% 90.85%
13 Z.AI GLM 5 88.46% 83.63% 91.23%
14 GPT-5.1 87.44% 87.20% 92.54%
15 Gemini 3.1 Pro (Preview) 87.20% 85.44% 94.37%
16 Qwen 3.5 Flash 87.13% 83.81% 86.38%
17 Qwen 3.5 122B 86.76% 83.02% 91.53%
18 MoonshotAI: Kimi K2.5 86.21% 81.35% 91.04%
19 GPT-5 Mini 85.94% 80.48% 92.62%
20 ByteDance Seed 1.6 85.62% 78.43% 90.70%
21 Writer: Palmyra X5 84.42% 83.95% 79.57%
22 DeepSeek V3 (2025-03-24) 83.31% 82.34% 81.99%
23 Mistral Large 2 83.10% 81.86% 82.41%
24 Claude Sonnet 4 82.84% 79.21% 88.72%
25 Claude Haiku 4.5 82.25% 78.96% 85.14%
26 Mistral Large 81.53% 82.02% 80.15%
27 Mistral Medium 3.1 81.51% 81.70% 77.83%
28 GPT-5.2 80.71% 80.36% 90.26%
29 Mistral Large 3 80.22% 81.21% 85.43%
30 Claude 3.5 Sonnet 80.13% 78.69% 84.24%
31 ByteDance Seed 1.6 Flash 79.96% 81.51% 73.27%
32 Grok 4.1 Fast 78.96% 82.14% 89.55%
33 o4 Mini 78.87% 82.04% 88.35%
34 Mistral Small Creative 78.30% 80.29% 73.27%
35 o4 Mini High 78.26% 82.72% 90.29%
36 Claude 3.5 Haiku 77.75% 75.28% 83.73%
37 WizardLM 2 8x22b 77.45% 79.06% 71.07%
38 Gemini 2.5 Pro 77.40% 81.03% 88.53%
39 GPT-4.1 76.96% 81.24% 88.68%
40 Aion 2.0 76.51% 80.24% 89.21%
41 Ministral 8B 76.29% 76.87% 64.87%
42 Ministral 3B 75.29% 75.49% 61.29%
43 Claude 3.7 Sonnet 75.00% 76.31% 83.39%
44 Ministral 3 14B 74.59% 79.11% 72.54%
45 Ministral 3 3B 74.49% 75.45% 67.22%
46 Z.AI GLM 4.6 74.32% 78.86% 89.11%
47 DeepSeek V3.2 73.33% 79.95% 82.25%
48 Z.AI GLM 4.5 73.22% 76.56% 86.27%
49 Llama 3.1 8B 73.18% 76.54% 63.37%
50 Ministral 3 8B 72.81% 77.26% 71.76%
51 Rocinante 12B 72.56% 81.94% 54.55%
52 LFM2 24B 72.22% 78.10% 58.77%
53 Z.AI GLM 4.7 71.01% 78.89% 88.69%
54 Llama 3.1 70B 70.76% 72.78% 78.40%
55 Gemini 2.5 Flash 70.64% 77.57% 80.60%
56 DeepSeek V3 (2024-12-26) 69.79% 77.88% 83.68%
57 Z.AI GLM 4.7 Flash 69.49% 77.36% 84.82%
58 Gemini 3 Flash (Preview, Reasoning) 69.48% 75.87% 90.50%
59 Grok 4 Fast 69.13% 77.03% 86.15%
60 Mistral Small 3.2 24B 68.93% 71.87% 78.60%
61 Hermes 3 405B 68.75% 80.92% 82.86%
62 Grok 4 68.65% 77.34% 88.12%
63 DeepSeek V3.1 68.61% 77.45% 82.39%
64 Mistral NeMO 67.00% 76.72% 65.04%
65 Gemini 3 Pro (Preview) 67.00% 77.77% 88.79%
66 DeepSeek-V2 Chat 66.95% 77.20% 84.83%
67 Gemini 2.5 Flash Lite (Reasoning) 65.73% 71.64% 85.75%
68 Gemini 2.5 Flash (Reasoning) 65.33% 76.30% 86.51%
69 Arcee AI: Trinity Large (Preview) 65.33% 75.26% 73.33%
70 GPT-4o, May 13th (temp=0) 64.81% 74.89% 85.36%
71 Gemini 3 Flash (Preview) 64.78% 75.04% 85.35%
72 Gemma 3 27B 64.44% 78.79% 77.85%
73 Qwen 2.5 72B 64.34% 75.16% 75.46%
74 Cohere Command R+ (Aug. 2024) 64.24% 77.70% 69.03%
75 Arcee AI: Trinity Mini 63.42% 74.01% 70.90%
76 GPT-4o, Aug. 6th (temp=0) 63.20% 73.65% 82.45%
77 Gemini 2.5 Flash Lite 62.67% 75.05% 81.08%
78 GPT-5 Nano 62.32% 67.04% 82.60%
79 Llama 3.1 Nemotron 70B 62.12% 71.71% 74.70%
80 Qwen 3.5 Plus (2026-02-15) 61.06% 77.07% 85.96%
81 Claude 3 Haiku 60.58% 74.53% 71.19%
82 GPT-4o, Aug. 6th (temp=1) 59.42% 75.50% 82.62%
83 Hermes 3 70B 59.25% 77.41% 72.57%
84 Stealth: Aurora Alpha 59.16% 67.54% 83.79%
85 GPT-4o, May 13th (temp=1) 58.51% 75.88% 83.80%
86 GPT-4.1 Mini 57.72% 74.52% 83.20%
87 GPT-4o Mini (temp=0) 56.91% 73.10% 78.29%
88 GPT-4.1 Nano 55.61% 71.81% 71.94%
89 Gemma 3 12B 54.87% 75.38% 78.41%
90 GPT-4o Mini (temp=1) 53.50% 74.37% 79.08%
91 Gemma 3 4B 52.95% 72.10% 68.57%