跳到正文
Hacker News 熱門(buzzing.cc 中文翻譯)·· 2 天前AI 評分30

Gemini 4 Neon 榮登《Models Engineering》指數榜首

Gemini 4 Neon 荣登《Models Engineering》指数榜首

AI 導讀

Gemini 4 Neon 在 Models Engineering Index 排名第一,得分為 68.0。該指數評估語言模型在工程任務上的實際表現。

正文

Mei—or, the Models Engineering Index—calculates language models' practical performance on engineering tasks. Mei calculates language models' engineering performance.

.

Performance versus .

Help spread the word with a GitHub star (opens in a new tab)

Rankings

Index bar
14 Neon Google68.01.6×0.83×
2Fable 5.2 Anthropic65.36.4×2.1×
36.2 Astra OpenAI61.51.3×0.29×
44 Carbon Google61.41.5×0.89×
5Opus 5.5 Anthropic61.13.4×1.9×
6Sonnet 5.5 Anthropic57.83.1×3.4×
74 Hydrogen Google56.91.1×1×
86.2 Sol OpenAI55.90.37×0.31×
94-Max Alibaba49.08.2×1.2×
10K3.1 Pro Moonshot48.41.1×0.74×
116.1 Terra OpenAI47.00.87×1.7×
12Wonder Full Meta46.41.5×0.51×
13Wonder Meta45.51.3×0.62×
14V4.1 Flash DeepSeek43.60.12×1.5×
15Haiku 5.5 Anthropic42.90.41×1.7×
16Grok 4.7 xAI41.33.1×0.8×
17GLM 5.3 Z.ai40.42×1.5×
18Hy4 Tencent37.20.34×1.2×
19V2.6-Pro Xiaomi35.30.12×0.96×
206 Luna OpenAI34.90.064×0.65×

Comparing and

來源:Hacker News 熱門(buzzing.cc 中文翻譯) · git-aarvin.github.io