跳过主要内容
评测来源/MCP Atlas
观察中专业工作

MCP Atlas

Scale AI · MCP Atlas 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「专业工作」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

Muse Spark 1.188.188.12026-11-05
Muse Spark 1.188.188.12026-07-09
Muse Spark 1.188.188.12026-10-18
Claude Opus 585.885.82026-10-18
Claude Opus 585.885.82026-11-05
GLM 5.384.284.22026-11-05
GLM 5.384.284.22026-10-18
Claude Fable 583.383.32026-10-18
Claude Fable 583.383.32026-11-05
Kimi K382.382.32026-10-18
Kimi K382.382.32026-11-05
Claude Opus 4.882.282.22026-11-05
Claude Opus 4.882.282.22026-10-18
GPT-5.6 Sol81.881.82026-11-05
GPT-5.6 Sol81.881.82026-10-18
Inkling Small79.279.22026-11-05
Inkling Small79.279.22026-10-18
Claude Opus 4.779.179.12026-11-05
Claude Opus 4.779.179.12026-10-18
Gemini 3.1 Pro Preview78.278.22026-11-05
Gemini 3.1 Pro Preview78.278.22026-05-19
Gemini 3.1 Pro Preview78.278.22026-10-18
Claude Opus 4.676.876.82026-10-18
Claude Opus 4.676.876.82026-11-05
Qwen3.7 Max76.476.42026-05-19
Kimi K2.7 Code7676.02026-06-12
Inkling7676.02026-11-05
Inkling7676.02026-10-18
GPT-5.575.375.32026-04-23
GPT-5.575.375.32026-10-18
GPT-5.575.375.32026-11-05
MiniMax-M374.274.22026-06-01
GPT-5.470.670.62026-11-05
GPT-5.470.670.62026-10-18
Gemini 3 Pro Preview70.370.32026-11-05
Gemini 3 Pro Preview70.370.32026-10-18
Claude Opus 4.569.869.82026-11-05
Claude Opus 4.569.869.82026-10-18
Claude Sonnet 4.669.569.52026-10-18
Claude Sonnet 4.669.569.52026-11-05
GPT-5.267.667.62026-11-05
GPT-5.267.667.62026-10-18
Gemini 3 Flash Preview6262.02026-11-05
Gemini 3 Flash Preview6262.02026-10-18
Claude Sonnet 4.559.559.52026-11-05
Claude Sonnet 4.559.559.52026-10-18
Gemini 3.1 Flash Lite57.157.12026-10-18
Gemini 3.1 Flash Lite57.157.12026-11-05
GPT-5.4 mini56.756.72026-10-18
GPT-5.4 mini56.756.72026-11-05
o3-pro44.544.52026-10-18
o3-pro44.544.52026-11-05
Claude Haiku 4.540.240.22026-10-18
Claude Haiku 4.540.240.22026-11-05

常见问题