# Toolathlon → MCP Atlas

Calibration · https://benchgap.net/calibration/1521 · data retrieved 2026-10-29

MCP Atlas is estimated from Toolathlon with a inverse Michaelis–Menten curve fitted on 17 models measured on both: y = 2.01861·(x − 0.0000) / (0.0000 + 2.0000 − x), R² = 0.58, cross-validated error 11.8 pp. It is used for 0 estimates.

## Fitted on

| Model | Toolathlon | MCP Atlas |
|---|---|---|
| [Claude Opus 4.5](https://benchgap.net/model/claude-opus-4-5.md) | 43.5% | 42.3% |
| [Claude Opus 4.8](https://benchgap.net/model/claude-opus-4-8.md) | 59.9% | 82.2% |
| [DeepSeek V4 Flash 0731](https://benchgap.net/model/deepseek-v4-flash-0731.md) | 47.8% | 69.0% |
| [DeepSeek V4 Pro 0813](https://benchgap.net/model/deepseek-v4-pro-0813.md) | 51.8% | 73.6% |
| [Gemini 3.5 Flash](https://benchgap.net/model/gemini-3-5-flash.md) | 56.5% | 83.6% |
| [GLM-5](https://benchgap.net/model/glm-5.md) | 38.0% | 31.1% |
| [GLM-5.2](https://benchgap.net/model/glm-5-2.md) | 48.2% | 76.8% |
| [GPT-5.4](https://benchgap.net/model/gpt-5-4.md) | 54.6% | 70.6% |
| [GPT-5.4 mini](https://benchgap.net/model/gpt-5-4-mini.md) | 42.9% | 57.7% |
| [GPT-5.4 nano](https://benchgap.net/model/gpt-5-4-nano.md) | 35.5% | 56.1% |
| [GPT-5.5](https://benchgap.net/model/gpt-5-5.md) | 55.6% | 75.3% |
| [Kimi K2.6](https://benchgap.net/model/kimi-2-6.md) | 50.0% | 55.9% |
| [Kimi K2.5](https://benchgap.net/model/kimi-k2-5.md) | 27.8% | 29.5% |
| [Muse Spark 1.1](https://benchgap.net/model/muse-spark-1-1.md) | 75.6% | 88.1% |
| [Qwen3.5 397B](https://benchgap.net/model/qwen3-5-397b.md) | 36.3% | 46.1% |
| [Qwen3.6-35B-A3B](https://benchgap.net/model/qwen3-6-35b-a3b.md) | 26.9% | 62.8% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 39.8% | 48.2% |