Benchmark
mcp-universe-function-call
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.
All agentic benchmarksScores from MCP-Universe.
- Models measured
- 18
- Scores recorded
- 19
- Sources
- 1
- IQ weight
- 2.36
Score distribution
- 25.3%
- Mean
- 24.7%
- Median
- 8.7%
- Std dev
- 15.6%
- 10th percentile
- 35.1%
- 90th percentile
- 12.1%
- Min
This benchmark has a weight of 2.36 for chat models, measured over 18 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro Preview· MCP-Universe | 44.6% |
| 2 | Grok 4.1 Fast· MCP-Universe | 40.7% |
| 3 | Claude Sonnet 4.5· MCP-Universe | 35.1% |
| 4 | Claude Sonnet 4· MCP-Universe | 32.9% |
| 5 | Grok 4 Fast· MCP-Universe | 32.5% |
| 6 | Claude Sonnet 4thinking· MCP-Universe | 31.6% |
| 7 | Claude Haiku 4.5· MCP-Universe | 26.4% |
| 8 | Kimi K2 Thinking· MCP-Universe | 26.4% |
| 9 | GPT OSS 120B· MCP-Universe | 25.5% |
| 10 | Grok Code Fast 1· MCP-Universe | 24.7% |
| 11 | GLM 4.6· MCP-Universe | 20.8% |
| 12 | DeepSeek V3.1· MCP-Universe | 19.9% |
| 13 | GLM 4.5· MCP-Universe | 19.9% |
| 14 | GPT 4.1· MCP-Universe | 19.9% |
| 15 | Kimi K2 0905· MCP-Universe | 19.5% |
| 16 | DeepSeek V3.1 Terminus· MCP-Universe | 18.6% |
| 17 | DeepSeek V3.2 Exp· MCP-Universe | 15.6% |
| 18 | GPT 4o· MCP-Universe | 13.4% |
| 19 | GLM 4.5 Air· MCP-Universe | 12.1% |