MCP Atlas — leaderboard

Evaluating real-world tool use through the Model Context Protocol (MCP).

Metric: Score (self-reported). Source: benchmarklist.com. Status: saturation imminent. 32 models tracked.

Top models

#ModelScore
1Gemini 3.5 Flash83.6
2Claude Fable 583.3
3Muse Spark82.2
4Claude Opus 4.779.1
5Gemini 3.1 Pro (Preview) (High)78.2
6Claude Opus 4.877.8
7GLM-5.276.8
8Claude Opus 4.6 (Max)76.8
9Qwen 3.7 Max (Max)76.4
10Kimi K2.7 Code76
11GLM-5.175.6
12GPT-5.5 (xHigh)75.3
13MiniMax-M374.2
14Qwen 3.6 Plus74.1
15DeepSeek V4 Pro73.6

Interactive version: theaggregate.ai/benchmark?slug=mcp-atlas · How the rankings work · Data refreshed daily, snapshot 2026-07-22.