MCP Atlas Public
Agents · 2025-09-19
Real-world tool-use benchmark across 36 MCP servers and 220 tools.
Top models (higher is better)
| Model | Score |
|---|---|
| Muse Spark 1.1 | 88.1 |
| Claude Opus 5 | 85.8 |
| Claude Fable 5 | 84.7 |
| Gemini 3.5 Flash | 83.6 |
| GPT-5.6 Sol | 83.6 |
| GLM-5.2 | 82.6 |
| Muse Spark | 82.2 |
| Opus 4.8 | 82.2 |
| Gemini 3.5 Flash-Lite | 79.8 |
| Opus 4.7 | 79.1 |
| Gemini 3.1 Pro Preview | 78.2 |
| GPT-5.6 Luna | 77.0 |
| GLM-5.1 | 75.6 |
| GPT-5.5 | 75.3 |
| Qwen3.5 397B A17B | 74.2 |
| Opus 4.5 | 69.8 |
| Sonnet 4.6 | 69.5 |
| DeepSeek-V4-Flash | 69.0 |
| Kimi K2.5 | 64.4 |
| Opus 4.6 | 62.7 |
| Sonnet 4.5 | 59.5 |
| GLM-4.7 | 58.1 |
| Gemini 3.1 Flash-Lite Preview | 57.1 |
| GPT-5.4 Nano | 56.1 |
| Gemini 3 Pro Preview | 54.1 |
| GPT-5.1 | 50.1 |
| MiniMax M2.7 | 49.4 |
| GPT-5 Mini | 47.6 |
| Nemotron 3 Ultra 550B A55B | 47.4 |
| o3-pro | 44.5 |
| Haiku 4.5 | 40.2 |
| Gemini 2.5 Pro | 8.8 |
| Gemini 2.5 Flash | 3.4 |