模型排行榜

以下是当前主流大语言模型在基准测试中的综合排名,以及成本与效率的关键对比数据:

排名模型名称综合指数相对成本相对Token效率
14 Neon (Google)69.51.9×1×
2Fable 5.2 (Anthropic)65.86.5×2.1×
3Opus 5.5 (Anthropic)63.13.5×1.7×
46.2 Astra (OpenAI)60.51.4×0.29×
54 Hydrogen (Google)57.21.2×1.1×
6Sonnet 5.5 (Anthropic)57.22.5×2.5×
76.2 Sol (OpenAI)56.90.4×0.35×
8Wonder Full (Meta)49.31.1×0.49×
9K3.1 Pro (Moonshot)49.01.2×0.74×
104-Max (Alibaba)49.07.3×1.2×
116.1 Terra (OpenAI)47.20.83×1.7×
12Wonder (Meta)46.41.3×0.63×
13V4.1 Flash (DeepSeek)44.10.13×1.4×
14Grok 4.7 (xAI)42.73.9×0.9×
15Haiku 5.5 (Anthropic)42.50.4×1.6×
16GLM 5.3 (Z.ai)42.02.2×1.5×
17Hy4 (Tencent)37.30.33×1.1×
18V2.6-Pro (Xiaomi)35.40.13×0.93×
196 Luna (OpenAI)34.80.065×0.62×

注:数据基于媒体基准测试整理。成本与Token效率为相对值,以排名第一的模型为基准(1×)。