benchgap
Calibration

AA-SciCode → React Native Evals

React Native Evals is estimated from AA-SciCode with a linear curve fitted on 6 models measured on both: y = 0.4248·x + 0.5543, R² = 0.56, cross-validated error 4.7 pp. It is used for 29 estimates.

Estimated modelAA-SciCodeReact Native EvalsSource
Claude Fable 5.267.7%84.2%estimated ± 4.7 pp, low confidence
Claude Opus 5.566.9%83.8%estimated ± 4.7 pp, low confidence
Muse Wonder Full62.6%82.0%estimated ± 4.7 pp, low confidence
Claude Sonnet 5.561.0%81.3%estimated ± 4.7 pp, low confidence
MiMo-V2.6-Pro60.9%81.3%estimated ± 4.7 pp, low confidence
Kimi K3.1 Pro60.8%81.3%estimated ± 4.7 pp, low confidence
Muse Wonder60.3%81.0%estimated ± 4.7 pp, low confidence
Step 5 Preview58.9%80.5%estimated ± 4.7 pp, low confidence
Muse Spark Open58.4%80.2%estimated ± 4.7 pp, medium confidence
GPT-6 Sol57.6%79.9%estimated ± 4.7 pp, medium confidence
Grok 4.757.4%79.8%estimated ± 4.7 pp, medium confidence
GPT-6.1 Terra55.1%78.8%estimated ± 4.7 pp, medium confidence
Claude Haiku 5.555.0%78.8%estimated ± 4.7 pp, medium confidence
GPT-6 Luna54.6%78.6%estimated ± 4.7 pp, medium confidence
GPT-6.1 Sol54.2%78.5%estimated ± 4.7 pp, medium confidence
Ling 3.1 Flash54.1%78.4%estimated ± 4.7 pp, medium confidence
MiniMax M3.1 Flash53.4%78.1%estimated ± 4.7 pp, medium confidence
DeepSeek V4.1 Flash 101751.9%77.5%estimated ± 4.7 pp, medium confidence
MiMo-V2.6-Flash51.3%77.2%estimated ± 4.7 pp, medium confidence
Ling 3.0 Flash VL44.2%74.2%estimated ± 4.7 pp, medium confidence
A.X K241.0%72.8%estimated ± 4.7 pp, medium confidence
Nemotron 3.5 Bolt 74B-A9B39.7%72.3%estimated ± 4.7 pp, medium confidence
DeepSeek V3 032439.0%72.0%estimated ± 4.7 pp, medium confidence
North Mini Code38.8%71.9%estimated ± 4.7 pp, medium confidence
Mercury 2.538.5%71.8%estimated ± 4.7 pp, medium confidence
Mistral Small 3.127.8%67.2%estimated ± 4.7 pp, low confidence
Solar Pro 325.5%66.3%estimated ± 4.7 pp, low confidence
Granite 4.2 3B25.3%66.2%estimated ± 4.7 pp, low confidence
Ling 3.0 Tiny24.2%65.7%estimated ± 4.7 pp, low confidence

Fitted on

ModelAA-SciCodeReact Native Evals
Gemini 3.1 Pro58.7%78.9%
Gemma 4 31B45.5%75.2%
GPT-5.555.8%84.7%
GPT-OSS 120B34.0%71.6%
GPT-OSS 20B38.9%71.0%
MiniMax M2.750.1%71.4%