Calibration

Toolathlon-Verified → Hermes Index

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap
License: https://creativecommons.org/licenses/by-nc/4.0/

Hermes Index is estimated from Toolathlon-Verified with a Hill curve fitted on 8 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.83984^6.00 + x^6.00), R² = 0.30, cross-validated error 11.3 pp. It is used for 0 estimates.

Fitted on

Model	Toolathlon-Verified	Hermes Index
Claude Opus 5.5	77.8%	63.3%
Claude Sonnet 5.5	77.8%	53.1%
GLM-5.3	73.0%	39.2%
GLM-5.3-Flash	78.4%	34.9%
Hy4 preview	74.1%	26.1%
Kimi K3	73.2%	39.2%
Qwen3.8 Max	72.5%	36.2%
Step 5 Preview	74.1%	32.8%