MCP-Universe: leaderboard
Benchmark for LLMs and agents using real-world Model Context Protocol servers across location navigation, repository management, finance, 3D design, browser automation, and web search tasks.
Metric: Overall Success Rate (imported). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Around January 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 (High) | 44.16 | #98 (GPT-5) |
| 2 | GPT-5 (Medium) | 43.72 | #98 (GPT-5) |
| 3 | Grok 4 | 33.33 | #181 |
| 4 | Claude Sonnet 4 (Thinking) | 30.3 | #199 (Claude Sonnet 4) |
| 5 | Claude Sonnet 4 | 29.44 | #199 |
| 6 | Claude Opus 4.1 | 29.44 | #140 |
| 7 | Grok 4 Fast | 27.27 | #254 |
| 8 | Grok Code Fast 1 | 26.41 | #337 |
| 9 | O3 (Medium) | 26.41 | #124 (O3) |
| 10 | GLM-4.6 | 25.97 | #252 |
| 11 | O4 Mini (Medium) | 25.97 | #180 (O4 Mini) |
| 12 | GLM-4.5 | 24.68 | #275 |
| 13 | Claude 3.7 Sonnet | 24.24 | #249 |
| 14 | Qwen 3 Coder 480B A35B Instruct | 22.94 | #315 |
| 15 | Gemini 2.5 Pro | 22.08 | #152 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mcp-universe · How It Works · Data refreshed daily, snapshot 2026-10-20.