跳过主要内容
评测来源/LiveBench 语言与指令
观察中知识与语言

LiveBench 语言与指令

LiveBench · LiveBench 语言与指令 的公开模型评测结果。

官方评测
来源说明

该来源测量模型在「知识与语言」领域的具体交付表现,原始成绩会和统一能力读数分开展示。

限制: 不同版本、推理设置和测试集可能导致分数变化,请结合数据时间阅读。

使用方式

参与排名的来源会影响能力分;参考、交叉参考和观察来源仅用于对比与核验。

本来源的模型成绩

Gemini 3.8 Flash84.14685784.12026-06-25
Claude Fable 582.16214382.22026-06-25
Gemini 3.1 Pro Preview81.78971481.82026-06-25
GPT-6 Astra81.51542981.52026-06-25
GPT-6.1 Sol80.99928681.02026-06-25
Claude Fable 5.180.06485780.12026-06-25
Muse Spark 1.380.05685780.12026-06-25
Gemini 3.6 Flash79.02285779.02026-06-25
GPT-5.6 Sol78.63357178.62026-06-25
GPT-5.577.85877.92026-06-25
Kimi K377.43357177.42026-06-25
Grok 4.777.35957177.42026-06-25
Muse Wonder77.27342977.32026-06-25
Grok 4.676.93671476.92026-06-25
Qwen3.8 Max76.48514376.52026-06-25
Qwen3.7 Max76.48328676.52026-06-25
Grok 4.576.35757176.42026-06-25
Muse Spark 1.276.14628676.12026-06-25
Claude Sonnet 5.576.05071476.12026-06-25
GPT-6 Sol75.74214375.72026-06-25
GPT-5.475.53828675.52026-06-25
Claude Opus 4.875.30171475.32026-06-25
Claude Opus 5.574.95642975.02026-06-25
DeepSeek V4.1 Flash74.81457174.82026-06-25
Claude Opus 574.44714374.42026-06-25
DeepSeek V4 Pro 081373.86028673.92026-06-25
GLM 5.373.82642973.82026-06-25
GPT-5.6 Terra72.44914372.42026-06-25
Nemotron 3 Ultra 550B A55B72.31657172.32026-06-25
Claude Opus 4.671.86557171.92026-06-25
Muse Spark 1.171.65371471.72026-06-25
Inkling71.53771.52026-06-25
Claude Opus 4.771.52742971.52026-06-25
DeepSeek V4 Flash 073171.37028671.42026-06-25
Claude Opus 4.570.56671470.62026-06-25
GPT-5.269.50142969.52026-06-25
Gemini 3.5 Flash Lite69.20171469.22026-06-25
DeepSeek V4 Pro69.11369.12026-06-25
Kimi K2.668.97957169.02026-06-25
Claude Sonnet 4.668.74157168.72026-06-25
Claude Sonnet 568.62071468.62026-06-25
Grok Build 0.168.32328668.32026-06-25
GLM-5.268.27028668.32026-06-25
Grok 4.367.39142967.42026-06-25
DeepSeek V4 Flash66.13171466.12026-06-25
MiniMax-M365.79142965.82026-06-25
Kimi K2.7 Code65.55485765.62026-06-25
Qwen3.6 Plus65.47657165.52026-06-25
GPT-5.6 Luna65.45565.52026-06-25
Claude Haiku 5.565.20265.22026-06-25
GPT-5.4 mini64.58157164.62026-06-25
GPT-6 Luna63.60228663.62026-06-25
GLM-5.3 Flash63.31563.32026-06-25
Mistral Large 458.31571458.32026-06-25
Qwen3.6 27B57.54757.52026-06-25

常见问题