List your company

Complex Multi-constraint Reasoning on MER-Bench Complex

802-P AccuracyGemini-3.0-Flash
37.8848.81559.7570.685Feb 7, 2026

Evaluation results

MethodLinks
Gemini-3.0-FlashType=Closed, Pointwise VIEScore setting=true2026.02805319.550.80.594
SpatialRewardType=Gen.2026.027849.521.548.30.549
Gemini-2.5-ProType=Closed, Pointwise VIEScore setting=true2026.027546.51746.20.546
GPT-5Type=Closed, Pointwise VIEScore setting=true2026.0272391642.30.472
EditRewardType=Disc.2026.02704915.544.80.49
GPT-4.1Type=Closed, Pointwise VIEScore setting=true2026.02662912.535.80.395
EditScore-8BType=Gen.2026.02613410350.393
Qwen3-VL-8BType=Gen.2026.0239.562.5160.182