List your company

ResearchFormerly WizWand

Open-ended Reasoning

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

private CCTV 80 videos (test)

39.47Rescaled BertScore F1

6

May 22, 2026

PPO w/ Structure Reward

57.05Creative Score

5

Mar 31, 2026

64.3Accuracy

2

Feb 26, 2026