# Toolathlon → ResearchClawBench

Calibration · https://benchgap.net/calibration/1524 · data retrieved 2026-10-29

ResearchClawBench is estimated from Toolathlon with a Michaelis–Menten curve fitted on 9 models measured on both: y = 0.2682·x / (0.25066 + x), R² = 0.40, cross-validated error 2.3 pp. It is used for 0 estimates.

## Fitted on

| Model | Toolathlon | ResearchClawBench |
|---|---|---|
| [Claude Opus 4.8](https://benchgap.net/model/claude-opus-4-8.md) | 59.9% | 21.1% |
| [Gemini 3.5 Flash](https://benchgap.net/model/gemini-3-5-flash.md) | 56.5% | 18.0% |
| [GLM-5.2](https://benchgap.net/model/glm-5-2.md) | 48.2% | 20.7% |
| [GPT-5.4](https://benchgap.net/model/gpt-5-4.md) | 54.6% | 15.3% |
| [GPT-5.5](https://benchgap.net/model/gpt-5-5.md) | 55.6% | 17.0% |
| [Kimi K2.6](https://benchgap.net/model/kimi-2-6.md) | 50.0% | 18.0% |
| [Kimi K2.5](https://benchgap.net/model/kimi-k2-5.md) | 27.8% | 14.0% |
| [Qwen3.5 397B](https://benchgap.net/model/qwen3-5-397b.md) | 36.3% | 14.2% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 39.8% | 18.0% |