# τ³-bench results → MCP-Tasks

Calibration · https://benchgap.net/calibration/1496 · data retrieved 2026-10-29

MCP-Tasks is estimated from τ³-bench results with a Hill curve fitted on 5 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.65083^6.00 + x^6.00), R² = 0.83, cross-validated error 3.6 pp. It is used for 6 estimates.

| Estimated model | τ³-bench results | MCP-Tasks | Source |
|---|---|---|---|
| [MiMo-V2.5-Pro](https://benchgap.net/model/mimo-v2-5-pro.md) | 72.9% | 79.7% | estimated ± 3.6 pp, low confidence |
| [Nemotron 3 Ultra](https://benchgap.net/model/nemotron-3-ultra.md) | 70.9% | 75.1% | estimated ± 3.6 pp, low confidence |
| [GLM-5.1](https://benchgap.net/model/glm-5-1.md) | 70.6% | 74.4% | estimated ± 3.6 pp, medium confidence |
| [MiniMax M2.7](https://benchgap.net/model/minimax-m2-7.md) | 67.6% | 66.8% | estimated ± 3.6 pp, medium confidence |
| [Pokee-Isaac 28B](https://benchgap.net/model/pokee-isaac-28b.md) | 66.2% | 63.1% | estimated ± 3.6 pp, medium confidence |
| [Granite 4.2 30B](https://benchgap.net/model/granite-4-2-30b.md) | 62.0% | 51.3% | estimated ± 3.6 pp, low confidence |

## Fitted on

| Model | τ³-bench results | MCP-Tasks |
|---|---|---|
| [Claude Opus 4.5](https://benchgap.net/model/claude-opus-4-5.md) | 70.2% | 71.8% |
| [GLM-5](https://benchgap.net/model/glm-5.md) | 65.6% | 60.8% |
| [Kimi K2.5](https://benchgap.net/model/kimi-k2-5.md) | 65.7% | 59.1% |
| [Qwen3.5 397B](https://benchgap.net/model/qwen3-5-397b.md) | 68.4% | 74.2% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 70.7% | 74.1% |