在“模型工程指数”(Mei)发布的最新排名中,Google推出的Gemini 4 Neon凭借在语言模型工程任务实际性能评估中的表现,以69.5的Index得分位居第一 1。该模型的成本比率为1.9×,Token消耗为1× 1。此次排名涵盖了Anthropic、OpenAI和Meta等多家主要科技公司的模型,并详细列出了各模型的指数得分、成本比率及Token消耗情况 1。
紧随其后的是Anthropic的两款模型,其中Fable 5.2以65.8分排名第二,Opus 5.5以63.1分位列第三 1。OpenAI的Astra 6.2则排在第四位,其Index得分为60.5 1。
Google’s Gemini 4 Neon has secured the top position on the latest Models Engineering Index (Mei), which evaluates language models based on their actual performance in engineering tasks 1. The model achieved an index score of 69.5, placing it ahead of competitors from major AI developers including Anthropic, OpenAI, and Meta 1. This ranking provides a comparative analysis of various models, detailing specific metrics such as index scores, cost ratios, and token consumption 1.
The index highlights efficiency metrics alongside performance, noting that Gemini 4 Neon recorded a cost ratio of 1.9× and a token consumption rate of 1× 1. Following closely in second place is Anthropic’s Fable 5.2, which scored 65.8 on the index 1. Anthropic also holds the third spot with its Opus 5.5 model, earning a score of 63.1, while OpenAI’s Astra 6.2 ranks fourth with a score of 60.5 1.
评论
还没有评论,欢迎留下第一条。