跳过主要内容
评测来源/CyberBench Patch v1.1
观察中编程

CyberBench Patch v1.1

Vals AI · CyberBench Patch v1.1 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「编程」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

GPT-6 Sol77.9878.02026-11-05
Gemini 4 Argon77.8677.92026-11-05
GPT-5.6 Sol76.3176.32026-11-05
GPT-6 Luna76.2576.32026-11-05
Claude Haiku 5.575.4875.52026-11-05
MiMo V2.6 Flash75.3675.42026-11-05
Kimi K375.2475.22026-11-05
DeepSeek V4.1 Flash73.6973.72026-11-05
GPT-5.6 Luna73.6373.62026-11-05
DeepSeek V4 Pro 081372.8672.92026-11-05
MiMo V2.6 Pro72.8672.92026-11-05
Muse Spark 1.372.7472.72026-11-05
GPT-5.6 Terra72.0872.12026-11-05
DeepSeek V4 Flash 073172.0872.12026-11-05
GLM 5.372.0872.12026-11-05
Grok 4.571.0171.02026-11-05
Claude Fable 5.170.4270.42026-11-05
Muse Spark 1.269.5269.52026-11-05
Grok 4.769.4669.52026-11-05
GLM-5.3 Flash67.8667.92026-11-05
Grok 4.667.6867.72026-11-05
Inkling66.7966.82026-11-05
Claude Opus 565.3665.42026-11-05
MiniMax-M36565.02026-11-05
Inkling Small64.1164.12026-11-05
Claude Sonnet 561.9161.92026-11-05
Claude Sonnet 5.559.5859.62026-11-05
Gemini 3.5 Flash Lite59.1159.12026-11-05
Claude Opus 5.555.3655.42026-11-05
Gemini 3.6 Flash45.3645.42026-11-05
Gemini 3.8 Flash43.7543.82026-11-05
GPT-6 Astra41.0741.12026-11-05
GPT-6.1 Sol39.2939.32026-11-05
Qwen3.8 Max28.5728.62026-11-05

常见问题