benchgap
Calibration

React Native Evals → AA Coding Index

AA Coding Index is estimated from React Native Evals with a Hill curve fitted on 9 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.80210^6.00 + x^6.00), R² = 0.72, cross-validated error 11.1 pp. It is used for 3 estimates.

Estimated modelReact Native EvalsAA Coding IndexSource
Composer 296.1%89.7%estimated ± 11.1 pp, low confidence
Composer 2 Fast94.9%87.9%estimated ± 11.1 pp, low confidence
DeepSeek V3.271.5%40.1%estimated ± 11.1 pp, low confidence

Fitted on

ModelReact Native EvalsAA Coding Index
Claude Opus 4.782.8%73.6%
Gemini 3.1 Pro78.9%68.8%
Gemma 4 31B75.2%43.4%
GPT-5.485.3%71.1%
GPT-5.584.7%74.9%
GPT-OSS 120B71.6%30.4%
GPT-OSS 20B71.0%20.7%
Kimi K2.577.2%46.8%
MiniMax M2.771.4%52.6%